Composites for Exploration Upper Stage

https://ntrs.nasa.gov/search.jsp?R=20160014739 2019-08-29T17:05:03+00:00Z

National Aeronautics and NASA/TM—2016–219433 Space Administration IS02 George C. Marshall Space Flight Center Huntsville, Alabama 35812

J.C. Fikes, J.R. Jackson, S.W. Richardson, and A.D. Thomas Marshall Space Flight Center, Huntsville, Alabama

T.O. Mann Langley Research Center, Hampton, Virginia

S.G. Miller Glenn Research Center, Cleveland, Ohio

December 2016 The NASA STI Program…in Profile

Since its founding, NASA has been dedicated to the • CONFERENCE PUBLICATION. Collected advancement of aeronautics and space science. The papers from scientific and technical conferences, NASA Scientific and Technical Information (STI) symposia, seminars, or other meetings sponsored Program Office plays a key part in helping NASA or cosponsored by NASA. maintain this important role. • SPECIAL PUBLICATION. Scientific, technical, The NASA STI Program Office is operated by or historical information from NASA programs, Langley Research Center, the lead center for projects, and mission, often concerned with NASA’s scientific and technical information. The subjects having substantial public interest. NASA STI Program Office provides access to the NASA STI Database, the largest collection of • TECHNICAL TRANSLATION. aeronautical and space science STI in the world. English-language translations of foreign The Program Office is also NASA’s institutional scientific and technical material pertinent to mechanism for disseminating the results of its NASA’s mission. research and development activities. These results are published by NASA in the NASA STI Report Specialized services that complement the STI Series, which includes the following report types: Program Office’s diverse offerings include creating custom thesauri, building customized databases, • TECHNICAL PUBLICATION. Reports of organizing and publishing research results…even completed research or a major significant providing videos. phase of research that present the results of NASA programs and include extensive data For more information about the NASA STI Program or theoretical analysis. Includes compilations Office, see the following: of significant scientific and technical data and information deemed to be of continuing • Access the NASA STI program home page at reference value. NASA’s counterpart of peer- reviewed formal professional papers but has less stringent limitations on manuscript length and • E-mail your question via the Internet to extent of graphic presentations.

• TECHNICAL MEMORANDUM. Scientific • Phone the NASA STI Help Desk at and technical findings that are preliminary or of 757 –864–9658 specialized interest, e.g., quick release reports, working papers, and bibliographies that contain • Write to: minimal annotation. Does not contain extensive NASA STI Information Desk analysis. Mail Stop 148 NASA Langley Research Center • CONTRACTOR REPORT. Scientific and Hampton, VA 23681–2199, USA technical findings by NASA-sponsored contractors and grantees. NASA/TM—2016–219433

Composites for Exploration Upper Stage

J.C. Fikes, J.R. Jackson, S.W. Richardson, and A.D. Thomas Marshall Space Flight Center, Huntsville, Alabama

T.O. Mann Langley Research Center, Hampton, Virginia

S.G. Miller Glenn Research Center, Cleveland, Ohio

National Aeronautics and Space Administration

Marshall Space Flight Center • Huntsville, Alabama 35812

December 2016

i Acknowledgments

The following personnel are acknowledged for their valuable GRC: contribution to actively running the Program Management Noel Nemeth, Structural Analysis office for this project: Abigail Rodriguez, Safety and Mission Assurance Lead Dan Scheiman, Panel Inspection NASA Marshall Space Flight Center (MSFC): Kenneth Smith, Panel Testing John Vickers, Project Manager John Thesken, Structural Analysis John Fikes, Deputy Project Manager Tiffany Williams, Panel Fabrication Justin Jackson, Project Lead Engineer Goddard Space Flight Center: Robert Gower, Scheduler Babak Farrokh, Structural Analysis Frank Ledbetter, Project Technical Support Kenneth Segal, Structural Design, Manufacturing & Risk Management LaRC: Lynn Machamer, Project Coordinator Darrell Branscome, Subject Matter Expert Dana Solomon, Budget Analyst Andrew Bergan, Structural Analysis Ruth Conrad, Safety & Mission Assurance Eric Burke, Non-Destructive Evaluation Deborah Cran, Chief Engineer Patrick A. Cosgrove, Project Lead Stephen Richardson, Analysis and Design Lead Mark Hillburger, Subject Matter Expert Allyson Thomas, Analysis and Design Co-Lead Kay Little, Program Analyst Larry Pelham, Manufacturing Lead Zoran Martinovic, Structural Analysis Erika Alvarez, Chief Engineer Joe McKenny, Structural Design, ISAAC Support NASA Glenn Research Center (GRC): Thuan Nguyen, ISAAC Programming Sandi Miller, Materials Lead Sverrir “Sev” Rosario, Structural Analysis NASA Langley Research Center (LaRC): Arunkumar Satyanarayana, Structural Analysis Troy Mann, Joint Structures Lead, Analysis Co-Lead Jamie Shiflett, Structural Design David Sleight, Structural Analysis The authors also acknowledge the following personnel for Brian Stewart, ISAAC Integration Manager significant contributions to this project: Chauncey Wu, ISAAC Lead GRC: MSFC: Bob Allen, Structural Analysis Lauren Badia John (Andy) Smith Eric Baker, Structural Analysis John Bausano Todd Swearingen Brett Bednarcyk, Structural Analysis Zenia Garcia Norris Vaughn Bill Brown, Panel Testing William (Chad) Hastings Stephen Wess Chris Burke, Panel Testing Alan Nettles Rob Wingate Cameron C. Cunningham, Analysis and Design Dawn R. Phillips Greg Zelinsky Mike Doherty, Project Lead Paula Heimann, Panel Fabrication NASA also acknowledges the following personnel from our Dan Kosareo, Design & Analysis major suppliers for their valuable contributions to this project: Ted Kovacevich, Structural Analysis Tom Krivanek, Analysis and Design Lead Janicki Industries Brad Lerch, Test & Evaluation Lead Cytec Industries Linda McCorkle, Panel Inspection Southern Research Institute (SRI) Dutch Myers, Structural Analysis NCAM

Available from:

NASA STI Information Desk Mail Stop 148 NASA Langley Research Center Hampton, VA 23681–2199, USA 757–864–9658

This report is also available in electronic form at

ii PREFACE

This Technical Memorandum documents the work accomplished by the NASA Composites for Exploration Upper Stage team from when the project received authority to proceed to systems requirements review (KDP-A) on June 6, 2014, to when the project received a cancellation notice from the Technology Demonstration Missions program office on February 4, 2016. The project was discontinued due to the FY 2016 NASA Appropriations Bill impacts to the Space Technology Mission Directorate budget. The initial project scope was to develop, build, and test two 8.4-m composite structures (forward and aft skirts) applicable to the Exploration Upper Stage for the Space Launch System vehicle. This technology project would mature the dry structure composite technologies to a Technology Readiness Level 6 at the 8.4-m-diameter scale. The project went through numerous scope trade studies, including applications to the Universal Stage Adapter as well as technology development accomplishments in the areas of materials, joints, design and manufacturing.

iii iv EXECUTIVE SUMMARY

The Composites for Exploration Upper Stage (CEUS) was a 3-year, level III project within the Technology Demonstration Missions program of the NASA Space Technology Mission Direc- torate. Studies have shown that composites provide important programmatic enhancements, including reduced weight to increase capability and accelerated expansion of exploration and science mission objectives. The CEUS project was focused on technologies that best advanced innovation, infusion, and broad applications for the inclusion of composites on future large human-rated launch vehicles and spacecraft. The benefits included near- and far-term opportunities for infusion (NASA, industry/commercial, Department of Defense), demonstrated critical technologies and technically implementable evolvable innovations, and sustained Agency experience.

The initial scope of the project was to advance technologies for large composite structures applicable to the Space Launch System (SLS) Exploration Upper Stage (EUS) by focusing on the affordability and technical performance of the EUS forward and aft skirts. The project was tasked to develop and demonstrate critical composite technologies with a focus on full-scale materials, design, manufacturing, and test using NASA in-house capabilities. This would have demonstrated a major advancement in confidence and matured the large-scale composite technology to a Tech- nology Readiness Level 6. This project would, therefore, have bridged the gap for providing composite application to SLS upgrades, enabling future exploration missions.

v vi TABLE OF CONTENTS

1. PROJECT OVERVIEW...... 1

1.1 Management Approach...... 3 1.2 Risk Management...... 4 1.3 Lessons Learned...... 6

2. MATERIALS ...... 14

2.1 Selection and Preexisting Data ...... 14 2.2 Prepreg Procurement Specification...... 15 2.3 Storage and Handling ...... 15 2.4 Film Adhesive Out-Time Study ...... 15 2.5 Equivalency Tests...... 16 2.6 Safety and Mission Assurance...... 27 2.7 Computational/Model Based Material...... 28 2.8 Digimat Test Simulation...... 32

3. DESIGN ...... 37

3.1 Composites for Exploration Upper Stage Skirt Design ...... 37 3.2 Wall Construction Trade Study...... 38 3.3 Design Requirements for Structural Test Article...... 42 3.4 Development of the Skirt Structural Test Article Layout ...... 43 3.5 Evaluation of Model Based Design Tools...... 46 3.6 Composite Universal Stage Adapter...... 51 3.7 Structural Test Article...... 52 3.8 Pathfinder Design Effort...... 56

4. JOINTS AND ANALYSIS OVERVIEW...... 58

4.1 Composites for Exploration Upper Stage Skirt Activities...... 58 4.2 Composites for Exploration Upper Stage Universal Stage Adapter Activities ...... 72

5. MANUFACTURING ...... 91

5.1 Tooling Development ...... 91 5.2 Composites for Exploration Upper Stage Tooling Requirements ...... 92 5.3 Assembly Concept ...... 96 5.4 Automated Fiber Placement Trials...... 98 5.5 Curing...... 105

vii TABLE OF CONTENTS (Continued)

6. TEST...... 117

7. SUMMARY...... 120

APPENDIX A—LESSONS LEARNED...... 121

APPENDIX B—MATERIALS ...... 126

B.1 8552 IM7 Unidirectional Prepeg Reports ...... 127

APPENDIX C—TECHNOLOGY READINESS LEVEL ASSESSMENT...... 237

APPENDIX D—DESIGN DATA SHEETS AND DRAWINGS ...... 239

APPENDIX E—MANUFACTURING ...... 256

REFERENCES ...... 266

viii LIST OF FIGURES

1. Project evaluation: (a) Scope options and (b) USA risk reduction evaluation ...... 3

2. Project organization structure ...... 4

3. Core and adhesive failure within the coupons fabricated from 20-day– aged film adhesive ...... 16

4. Recommended bagging sequence...... 17

5. LaRC ISAAC system...... 18

6. Equivalency panel fabrication at LaRC...... 19

7. Panel layups tested on CGtech VCP...... 19

8. CGTech VERICUT tested virtually ...... 20

9. C-scan images of the 16-ply panels fabricated for (a) UNT, (b) UNC, and (c) OHT tests ...... 20

10. Photomicrographs of GRC panels: (a) 24 ply and (b) 36 ply, LaRC panels: (c) 24 ply and (d) 36 ply, and MSFC panels: (e) 24 ply and (f) 36 ply ...... 21

11. Digimat platform and available modulus a GSFC ...... 29

12. Design and material property computation of a unitape material ...... 30

13. A 3D woven RVE creation and corresponding FE mesh ...... 31

14. FEA results and property extraction for a 3D woven RVE ...... 31

15. Test matrix definition...... 32

16. Test variability definition...... 33

17. Simulation and allowable computation ...... 34

18. Digimat software verification approach...... 35

ix LIST OF FIGURES (Continued)

19. RVE: (a) With 60% FV and (b) with 60% FV and 7% void content ...... 36

20. Early concept of a cryogenic Upper Stage for the SLS...... 37

21. Composites for Upper Exploration Stage CEUS-TRADE-001...... 39

22. DDS for the forward skirt STA ...... 43

23. Forward skirt STA layout...... 45

24. Forward skirt STA layout second angle ...... 45

25. CREO model of tool surface ...... 47

26. Fibersim is used to generate plies ...... 47

27. Fibersim generated 3D cross section of laminate showing interleaved plies...... 48

28. Fibersim has a ply sequence editor that allows the designer to easily edit the ply sequence...... 48

29. CGTech VERICUT VCPe with imported Creo model and Fibersim ply data...... 49

30. CGTech VCS manufacturing simulation...... 50

31. CGTech VERICUT VCPe-generated course data for AFP (reinforcement ply) ...... 50

32. CUSA2 STA traded assembly options ...... 51

33. Interface definition orf the STA assembly fixture (sheet 1)...... 53

34. Interface definition orf the STA assembly fixture (sheet 2)...... 53

35. Layout drawing for the STA (sheet 1) ...... 54

36. Layout drawing for the STA (sheet 2) ...... 55

37. Layout drawing for the STA (sheet 3) ...... 55

38. Pathfinder conceptual model...... 56

39. OML trade study considerations: (a) Two-part cone and cylinder and (b) alternate geometries ...... 57

x LIST OF FIGURES (Continued)

40. Aft skirt sizing model ...... 59

41. Joint interface details ...... 59

42. Configuration used for analysis of the critical size of kissing bonds at the facesheet-core interface ...... 63

43. Typical mesh used for the critical disband size analysis: (a) Front and (b) side ...... 64

44. Boundary conditions for the critical disband size analysis ...... 65

45. Typical first mode of eigenvalue buckling analysis ...... 66

46. Damage, load-displacement response, and out-of-plane deformation predicted using cohesive elements: (a) Damage propagation, (b) load displacement response, and (c) uz (in) ...... 67

47. Damage, load-displacement response, and out-of-plane deformation predicted using VCCT: (a) Damage propagation, (b) load displacement response, and (c) uz (in)...... 67

48. Critical disbond size ...... 69

49. Schematic of the ELC ...... 70

50. Wide-area, vacuum-assisted disbond inspection method ...... 71

51. Detectable size for vacuum-assisted disbond inspection ...... 71

52. USA options ...... 72

53. Analysis model for USA initial design concept ...... 73

54. USA: (a) Baseline geometry and (b) representative buckling results ...... 74

55. STA model with typical buckling eigenmode response ...... 76

56. STA core depth study results ...... 77

57. Simulator capability memorandum ...... 78

58. Full stack model—assembly of STA, simulators, loading rings, and load jacks ...... 79

xi LIST OF FIGURES (Continued)

59. Load versus displacement curve and radial deformation contours in stack ...... 79

60. Radial deformations in LH2 simulator: (a) Just before buckling and (b) at the end of the analysis ...... 80

61. Axial displacement of load jacks (in inches) ...... 80

62. Load versus displacement curve and radial deformation contours in stack ...... 81

63. Radial deformation contours in the STA: (a) Just before buckling and (b) just after buckling ...... 81

64. Strain contours in LH2 simulator: (a) Just before buckling and (b) just after buckling ...... 82

65. Displacement and load configurations (in inches) ...... 82

66. Load versus displacement curve and radial deformation contours in stack ...... 83

67. Radial deformation contours in the STA: (a) Just before buckling and (b) just after buckling ...... 83

68. Nodal imperfections applied to the STA ...... 84

69. Load deflection curve of transient dynamic nonlinear analysis ...... 85

70. Radial deformation of perfect STA model: (a) At buckling load and (b) at last load increment in the transient dynamic analysis ...... 85

71. Radial deformation of STA model with imperfections: (a) At buckling load and (b) at last load increment in the transient dynamic analysis ...... 86

72. Buckling and EWC FEMs with end-fitting details...... 88

73. Buckling performance: (a) 4.5 pcf and (b) 3.1 pcf core models ...... 89

74. Facesheet failure indices at buckling ...... 90

75. CEUS tooling design concept ...... 92

76. CEUS final tool...... 94

77. CEUS metrology (dimensions are in mm) ...... 94

xii LIST OF FIGURES (Continued)

78. CEUS tool installed in MSFC AFP cell ...... 95

79. CEUS tool AFP trials ...... 95

80. Assembly fixture concept and finished CEUS...... 98

81. Debulk vacuum bag schematic...... 99

82. AFP of first skin...... 100

83. Film adhesive installed on first composite skin...... 100

84. Honeycomb core cell irregular shapes...... 101

85. Foaming core splice installation...... 101

86. Paste core splice adhesive injection ...... 102

87. Second film adhesive layer with overlap ...... 103

88. View of honeycomb core after fiber placement of top skin, showing no damage ...... 103

89. AFP of second skin onto surface of honeycomb core...... 104

90. Autoclave cure vacuum bag ...... 104

91. Part loaded into autoclave ...... 105

92. Voids ...... 107

93. Tan delta peak at various 285 ºF hold times ...... 109

94. Ramp rate of 2 ºF/min with 30-min, 285 ºF hold time ...... 110

95. Ramp rate of 5 ºF/min with 30-min, 285 ºF/min to cure temperature ...... 111

96. Micrograph of 8552-1 DDS/TGDDM resin (darker section) deforming the Cytec FM-300 film adhesive (lighter section) at core interface ...... 112

97. Loss and storage modulus values of FM300-3 ...... 113

xiii LIST OF FIGURES (Continued)

98. Combining the results of the FM300-2 film adhesive and 8552-1...... 114

99. Clean cross section using developed cure cycle ...... 115

100. Typical profile of porosity using developed cure cycle...... 115

101. Example of void most likely caused from air entrainment ...... 115

102. Example of digital image correlation data: (a) Superimposed on a barrel that has been painted with a high contrast speckle pattern compared to (b) an analytical model of predicted displacement ...... 117

103. Test configuration in test stand 4699...... 118

104. Progress made in composite technologies ...... 120

xiv LIST OF TABLES

1. Level 1 requirements for CEUS-1 through CEUS-7 ...... 2

2. Open moderate risks at project termination ...... 5

3. Historic lessons learned March 2015 ...... 7

4. Ares I Interstage mapping to MSFC-RQMT-3479 ...... 9

5. Ares I First Stage frustum mapping to MSFC-RQMT-3479 ...... 10

6. Space Shuttle ET Fwd GH2 Press Line Fairing mapping to MSFC-RQMT-3479 ...... 10

7. Space Shuttle ET Intertank Access Door mapping to MSFC-RQMT-3479 ...... 11

8. Space Shuttle ET composite nose cone mapping to MSFC-RQMT-3479 ...... 12

9. Space Shuttle FWC mapping to MSFC-RQMT-3479 ...... 12

10. HST SLIC mapping to MSFC-RQMT-3479 ...... 13

11. Orion MPCV mapping to MSFC-RQMT-3479 ...... 13

12. Test matrix ...... 22

13. Lamina tensile strength and modulus data ...... 24

14. Unnotched tensile strength and modulus data ...... 24

15. Unnotched compression strength and modulus data ...... 25

16. Laminate short beam shear data ...... 25

17. Open-hole tension data ...... 26

18. Open-hole compression data ...... 26

19. OHT data ...... 40

xv LIST OF TABLES (Continued)

20. OHC data...... 41

21. Wall construction trade study mass results ...... 60

22. Buckling trade study results ...... 75

23. Summary of buckling load line load and principal strain ...... 84

24. Summary of buckling load sensitivity to geometric imperfections study ...... 86

25. Lightweight core validation test plan ...... 87

26 Predicted loads and end-shortening: Core trade study...... 89

27. Fiber (IM7) test matrix ...... 126

28. Matrix (8552-1 neat resin) test matrix...... 126

29. IM7/8552-1 lamina/laminate test matrix...... 126

30. Assessment of TRL/MRL of 8.4-m-diameter composite technologies...... 237

31. Assessment of composite technologies versus perceived impediments...... 238

xvi LIST OF ACRONYMS, SYMBOLS, AND ABBREVIATIONS

AFP automated fiber placement Al aluminum ANOVA analysis of variance ASTM American Society of Testing and Materials CAD computer-aided design CE chief engineer CEUS Composite Exploration Upper Stage CMM coordinate measuring machine CPT cured ply thickness CUSA composite for the universal stage adapter CV coefficient of variation DDS design data sheet DDT&E design, development, test, and evaluation DSC Differential Scanning Calorimetry (test) DTA damage threat assessment EFT-1 Exploration Flight Test 1 ELC elasticity laminate checker EOP edge of port ET external tank ETW elevated temperature wet EUS Exploration Upper Stage EWC edge-wise compression FCB fraction control board FE finite element

xvii LIST OF ACRONYMS, SYMBOLS, AND ABBREVIATIONS (Continued)

FEA finite element analysis FEM finite element model FEP fluorinated ethylene propylene FI failure index FS factors of safety FV fiber olumev FWC filament oundw case

GH2 gaseous hydrogen GRC Glenn Research Center GSFC Goddard Space Flight Center HEOMD Human Exploration and Operations Mission Directorate HST Hubble Space Telescope IDPP impact damage protection plan ISSAC integrated structual assembly of advanced composites ITAR International Traffic in Arms Regulation LaRC Langley Research Center

LH2 liquid hydrogen LOX liquid oxygen LSSM lightweight spacecraft structures and materials MPCV multipurpose crew vehicle MPR Marshall Procedural Requirements MRL manufacturing readiness assessment MSA MPCV stage adapter MSFC Marshall Space Flight Center NASA National Aeronautics and Space Administration NCAMP National Center for Advanced Materials Performance

xviii LIST OF ACRONYMS, SYMBOLS, AND ABBREVIATIONS (Continued)

NDE nondestructive evaluation NIAR National Institute of Aviation Research OHC open-hole compression OHT open-hole tension OML outer mold line OOA out-of-autoclave PDR preliminary design review PM project manager PMT project management team psf pounds per square foot RMS root-mean-square RTD room temperature dry RVE representative volume element SBKF shell buckling knockdown factor SBS short beam shear SLIC super-lightweight interchangeable carrier SLS Space Launch System SMA safety and mission assurance SRB solid rocket booster SSR systems requirements review STA structural test article STMD Space Technology Mission Directorate TPS Thermal Protection System TRL technology readiness level USA universal stage adapter VCCT virtual crack closure technique

xix LIST OF ACRONYMS, SYMBOLS, AND ABBREVIATIONS (Continued)

VCP VERICUT for Composites Programming

VCPe VERICUT for Composite Paths for Engineering

VCS VERICUT for Composites Simulation

VFM virtual flight model

WBS Work Breakdown Structure

xx NOMENCLATURE

D diameter

E modulus of elasticity

E ′ storage modulus E″ loss modulus Fz avail load force (z direction) h height I damage initiation

My moment (y direction) N load displacement nonlinearity

Ny loading direction P peak load

Tg transition

tc core thickness

tf facesheet thickness w width x location on x-axis y location on y-axis z location on z-axis

d loss tangent η viscosity

xxi xxii TECHNICAL MEMORANDUM

COMPOSITES FOR EXPLORATION UPPER STAGE

1. PROJECT OVERVIEW

The Composite for Exploration Upper Stage (CEUS) project technical scope was developed to (1) reduce technical impediments for the use of composites in launch vehicles and other NASA missions; (2) bridge the gap between technology demonstration and the final infusion into NASA missions; (3) streamline the design and manufacturing processes for launch vehicle applications to reduce development costs; and (4) address the use of conservative knockdown factors and limiting damage tolerance techniques. This scope was organized into the following four primary areas: materials, structures, manufacturing, and testing. The initial activity content was focused on the liquid hydrogen (LH2) tank forward and aft skirt of the Exploration Upper Stage (EUS). Level 1 requirements were generated for the initial CEUS technical scope (shown in table 1) and were updated during the project scope trade studies. The tasks for the initial scope included the following:

(1) Accelerated building block approach: • Coupon program. • Joint development (out-of-autoclave (OOA) jointing activity with composite joints). • Tool design and fabrication.

(2) Eight-segment forward LH2 skirt and aft LH2 skirt: • Fit in 20-ft autoclave. • Metallic joints (optional). • Analyze and test critical subcomponent specimens to develop and validate the composite design database.

(3) Design in a model-based environment: • Failure modes and loads predicted within 5% of measured values.

(4) Modeling and simulation included in manufacturing flow.

(5) Assessment of structural secondary attachments.

(6) Thermal testing:

(7) Multipurpose tool for LH2 forward skirt and LH2 aft skirt (outer mold line (OML) structures).

1 Table 1. Level 1 requirements for CEUS-1 through CEUS-7.

Structure Level 1 Requirements* CEUS-1 The CEUS project shall fabricate at least one full-scale technology demonstration unit Space Launch System (SLS)/EUS hydrogen tank skirt assembly. CEUS-2 The assembly shall include interfaces representative of those in the SRR SLS/EUS baseline design. CEUS-3 The assembly shall demonstrate structural integrity when exposed to qualification-level loads and environments representative of the anticipated SLS/EUS-induced and natural environments. CEUS-4 The assembly, including planned penetrations and interfaces, shall be at least 20% less massive than an analogous metallic skirt. CEUS-5 The CEUS project shall document design, analysis, manufacturing, inspection, and test data relevant to qualiftying large composite structures and infusing them into future human exploration vehicles. CEUS-6 The CEUS project shall provide estimated DDT&E costs for producing flight-qualified composite skirts. CEUS-7 The CEUS project shall use DDT&E methodologies consistent with SLS/EUS composite structure certification requirements.

•All CEUS level 1 requirements were approved.

(8) Forward skirt and aft skirt testing in a relevant environment.

(9) Forward skirt and aft skirt development for qualification.

(10) Publically disseminate all results.

The project successfully completed the Systems Requirements Review (SRR) for the initial technical scope. Several scope trade studies were accomplished after SRR that assessed deleting the aft skirt and adding additional large panel testing, and assessed technical scope related to the universal stage adapter (USA) instead of the LH2 skirts. These trades were either complete or nearing completion when the project was cancelled. Figure 1(a) depicts the scope options the project evaluated with the EUS LH2 skirts, and (b) depicts the USA risk reduction.

2 LH Forward 2 Coupon & Skirt Joint Testing LH2 Tank Structural LH2 Aft Skirt Concepts, Design, & Intertank Analysis

LOX Tank Advanced Manufacturing Thrust Manufacturing Structure Analysis & Simulation

Interstage Test/ Analysis & Correlation (a) (b)

F1_1624 Figure 1. Project evaluation: (a) Scope options and (b) USA risk reduction evaluation.

1.1 Management Approach

NASA’s Glenn Research Center (GRC), Langley Research Center (LaRC), and Marshall Space Flight Center (MSFC) formed a strong technical and management team to support the CEUS project. Collaborative partnership arrangements were made between the three NASA Centers to ensure the success of the project. Project team members were carefully selected based on previous experience and suitability to help the project reach its objectives and specific aims. GRC was responsible for leading the Materials Work Breakdown Structure (WBS) element, LaRC was responsible for leading the Joint Development and Structural Analysis WBS elements, and MSFC was responsible for design, manufacturing, testing, and project management. All Centers were responsible for the work in their lead roles, providing support roles in other WBS areas, as well as insight/oversight work with the in-house and contractor activities, developing requirements, and performing reviews. The organization chart in figure 2 indicates the LaRC, GRC, and MSFC roles in the project organization. The organization structure was designed to establish lines of responsibility and management accountability. Core management came from within MSFC to create a centralized project management team (PMT). The PMT consisted of the project manager (PM), the deputy project manager, and the chief engineer (CE). Project resource administrators (RAs) at each Center managed the business side of their Center’s tasks. The PM’s support staff was drawn from MSFC’s directorates and provided the PM with effective project control, permitted delegation of authority and responsibility, ensured short lines of communication and reporting, and permitted corrective actions to be taken at the proper level of responsibility.

3 J. Vickers–Project Manager Partnerships: J. Fikes–Deputy Project Manager Potential Leveraging: HEOMD–AES D. Crane, E. Alvarez–Chief Engineers Government Agencies, Industry, HEOMD–SLS J. Jackson–Project Engineer Academia, International MSFC

1. 2. 3. 4. 5. Project Management Materials Joint Development Design Manufacturing MSFC GRC LaRC MSFC MSFC John Vickers Sandi Miller Troy Mann Stephen Richardson Larry Pelham John Fikes Allyson Thomas

6. 7. 8. 9. Structural Analysis Testing & Evaluation Tool Design & Support LaRC MSFC Manufacture Contractors Troy Mann Dee Vancleave Contract Andy Finchum

Figure 2. Project organization structure. F2_1624

1.2 Risk Management

The CEUS project managed risk according to tailored requirements outlined in MPR 7120.1, Rev. G, “MSFC Engineering and Program/Project Management Requirements,” dated August 26, 2014.1 The Risk Management process consisted of the following:

• Identification of risk contributors. • Analyses to estimate probability and consequences. • Planning of risk mitigation. • Tracking to performance measures. • Controlling risk through adjustments to plans and control measures. • Communication of risk management activity. • Documentation throughout the process.

4 Risk was evaluated on a 5×5 matrix of likelihood and consequence, and the PM, CE, and chief safety officer had the authority to determine risk items to be entered in the system and to adjust the likelihood and consequence levels. The project assigned a risk owner for each risk item. In addition to the 5×5 assessment, each risk owner prepared and presented the tasks, funding, and schedule required to mitigate the risk and the impacts of not mitigating (technical, cost, schedule, safety).

At the time of termination, the project had eleven open risks, of which none were high (red), six were moderate (yellow), and five were low (green). As a result of several trade studies that kept the project in a transient state for several months prior to termination, several of the risks were still being formulated. As a result, many mitigation plans and risk statements were still to be defined at the time of termination. The open moderate risks are shown in table 2.

Table 2. Open moderate risks at project termination.

Risk Likelihood Consequence Risk Mitigation Impact of Not Mitigating If the MSFC 20-ft autoclave should fail, the CEUS schedule CEUS-01 2×5 Autoclave failure Watch may be delayed by up to 6 months into FY 2018. Given the current design criteria for composites, design decisions could lead the effort to overly conservative solutions that CEUS-14 3×3 Technical design criteria Mitigate will cost time, budget, and performance with no significant reduction in risk and not taking full advantage of potential savings.

Given that major SLS structural tests and test related activities CEUS-22 2×5 SLS test failure Watch will be performed concurrently at MSFC, a test failure in an adjacent test stand could damage the CEUS test article.

Manufacturing processes for composite structures typically Scale up—large composite do not lend themselves to linear scaling. As a result, there is CEUS-105 3×3 Mitigate structures inherent risk with the scaling of structures. Unforeseen issues that could arise may impact both cost and schedule.

CEUS-108 3×2 OOA joint manufacturing Mitigate Demonstration of large-scale OOA joint manufacturing.

CEUS-111 2×4 Assembly fixture—end ring Mitigate End ring for assembly fixture is schedule critical.

5 1.3 Lessons Learned

1.3.1 Review of Historical Lessons Learned

A review of the NASA Lessons Learned database and of previous composites projects was conducted. The purpose of the review was to capture lessons learned from prior Agency efforts and to build a plan to address those lessons specifically for the CEUS project. The information reviewed included the following:

• Composite Crew Module – NESC-RP-06-019, series of seven NASA Technical Memoranda (NASA/TM—2011–217185 through NASA/TM—2011–217191) detailing primary structure, design, materials, and processes, analysis, manufacturing, test, and nondestructive evaluation (NDE).

• Final Report of the X-33 Liquid Hydrogen Tank Test Investigation team (no document number assigned).

• Ares 1 Interstage – MSFC/EV31 internal documentation.

• Composite Cryotank – MSFC/EV31 internal documentation.

• SRB Composite Nose Cap Project – MSFC strength analysis group internal documentation.

• Composite Interstage Structural Concepts for Heavy Lift Launch Vehicles – Lightweight Spacecraft Structures & Materials (LSSM) project internal documentation.

• Composite Chronicles: A Study of the Lessons Learned in the Development, Production, and Service of Composite Structures – NASA-CR-4620.

• Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings, and Payload Attach Structures – Boeing internal documentation.

• NASA Preferred Reliability Practices – NASA-TM-4322A.

• NASA Lessons Learned Information System – Various lessons learned found using keyword composites.

The information was collected in a spreadsheet (app. A). Each lesson learned was entered in its own row with document number, document name, date, LLIS# (if applicable), statement of the lesson learned, applicability of the lesson to the CEUS project, and the plan to address the lesson in the CEUS project. A total of 94 lessons learned were considered for applicability to the CEUS project. The lessons learned were reviewed during the design and analysis meetings held for this project. The project was cancelled before a plan for every lesson could be formulated. Historic lessons learned are listed in table 3.

6 Table 3. Historic lessons learned March 2015.

Document Number Document Name Date Lesson Learned Applicable? Plan NASA/TM—2011– Composite Crew Nov. 2011 F-1: Many of the design, analysis, Yes The CEUS project is aware that there 217185; NESC- Module: Primary materials, and manufacturing les- can be scaling issues in manufactur- RP-06-019 Structure sons learned were not evident in the ing and analysis. The project is going coupon, element, or subcomponent to build a full-scale test article for final testing and only became evident verification of the design and when building the full-scale assembly. manufacturing.

NASA/TM—2011– Composite Crew Nov. 2011 F-5: Manufacturability, symmetry, and No No flight structure is ever perfectly opti- 217188; NESC- Module: Analysis other derived requirements resulted in mized. All engineering designs are RP-06-019 a CCM structure that does not have a compromise to meet sometimes zero MS everywhere. These derived conflicting requirements. requirements can lead to higher MS, thus higher mass than expected.

NASA/TM—2011– Composite Crew Nov. 2011 F-11: The need for repair was caused Yes Take special care with tooling (tethers?) 217189; NESC- Module: by a variety of sources including to avoid impact damage. Make sure RP-06-019 Manufacturing inadvertent impact damage, failed procedures for curing are followed cor- secondary cures, and secondary rectly. Take special care with the place- mechanical fastener hole ment of fastener holes; designers, make misalignment. sure to accommodate existing location constraints; manufacturing, make sure to follow drawings (STA design and manufacturing) NASA/TM—2011– Composite Crew Nov. 2011 F-1: Real-time monitoring of test Yes Generate pretest predictions and have 217190; NESC- Module: Test results against analytical predictions analysts at stress stations during test RP-06-019 was central to the success of the full- (STA analysis and test). scale test program. This enabled the team to push the applied load limits progressively while minimizing the risk of catastrophic failure.

1.3.2 Review of Previous Composites Program Fracture Control Plans and Certification Strategies

A review of previous NASA composites projects for human-rated hardware was conducted. The purpose of the review was to compare the approaches the projects used to satisfy fracture control and damage tolerance requirements and compare those to the intent of the current Agency damage tolerance standards. The projects reviewed were as follows:

• Ares I Interstage.

• Ares I First Stage Frustum.

• Space Shuttle External Tank (ET) Fwd Gaseous Hydrogen (GH2) Press Line Fairing. • Space Shuttle ET Intertank Access Door.

• Space Shuttle ET Composite Nose Cone.

• Space Shuttle Filament Wound Case (FWC).

7 • Hubble Space Telescope (HST) Super Lightweight Interchangeable Carrier (SLIC).

• Orion Multi-Purpose Crew Vehicle (MPCV).

Of the projects in this list, Ares I and the Space Shuttle FWC were cancelled for programmatic reasons prior to first flight. The remaining Shuttle and Hubble structures were successfully flown, and, at the time of writing this Technical Memorandum, Orion is still in development. Additionally, the SLS Block 1 MPCV Stage Adapter (MSA) composite diaphragm (flown on Orion Exploration Flight Test 1 (EFT-1) and scheduled to fly on SLS Exploration Mission 1 (EM-1)) and the Space Shuttle payload bay doors were being studied for this effort, but the CEUS project was overcome by events before the summaries could be made available to the team for review.

Fracture control requirements are provided in NASA-STD-5019, “Fracture Control Requirements for Spaceflight Hardware.” NASA-STD-5019 points to MSFC-RQMT-3479, “Fracture Control Requirements for Composite and Bonded Vehicle and Payload Structures,” for fracture control of composite structures.2 At the time of this study, the Agency damage tolerance requirements for fracture critical composite structures were specified in NASA-STD-5019, which invoked MSFC-RQMT-3479.3 Subsequent to this study, Revision A to NASA-STD-5019 was released, which contained all mandatory composite damage tolerance requirements directly in the standard in lieu of invoking MSFC-RQMT-3479. The expectation is that MSFC-RQMT-3479 will become inactive for new design, and fracture critical composite structures on future projects will have to satisfy the damage tolerance requirements of NASA-STD-5019, Rev A.

As stated in MSFC-RQMT-3479, the damage tolerant approach is the preferred approach for accepting fracture critical parts. In the event the required steps cannot be accomplished for a specific structure or joint, the developer may propose an alternate approach to the NASA Project Office through the fracture control plan. The Responsible Fracture Control Board (FCB) reviews the technical adequacy of the plan for the Project Office.

Fracture control of composite structures using the damage tolerant approach shall meet the steps listed below. Note that the damage tolerant approach does include a proof/acceptance test of the flight article. The following steps are the minimum required:

(1) Damage threat assessment (DTA). (2) Impact damage protection plan (IDPP). (3) Damage tolerant coupon tests. (4) Damage tolerant development tests. (5) Analytical support. (6) Damage tolerant full-scale component tests. (7) Implement impact damage protection plan. (8) NDE parts. (9) Acceptance proof test to 1.05 minimum. (10) Post-proof NDE. (11) In-service inspection.

8 MSFC-RQMT-3479 was released on June 29, 2006. Thus, MSFC-RQMT-3479 did not exist at the time of composite hardware development for projects prior to 2006, but the principal concepts of composite damage tolerance existed, and each project was required to meet fracture control for composite/bonded structures as shown in tables 4–11 in sections 1.3.2.1 through 1.3.2.8. In this summary, these projects were assessed for whether their activities met the intent of MSFC-RQMT-3479. The summaries were put together by members of the CEUS team as part of researching background information; i.e., the summaries were not performed by any agency-char- tered FCB.

In tables 4–11, each of the projects reviewed is mapped to the requirements in MSFC- RQMT-3479. Each of these summaries provides objective evidence to demonstrate successful certification of composite flight structure. It should be noted that for every project, the damage tolerance plan was vetted by an FCB and had Board approval.

1.3.2.1 Ares I Interstage. Design work for the Ares I Interstage began c. 2008. MSFC- RQMT-3479 was levied on the project via NASA-STD-5019. The fracture control plan was approved by the MSFC FCB on June 10, 2008, in memo EM20-08-FCB-014.

Table 4. Ares I Interstage mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA In work at time of cancellation IDPP Yes Damage tolerant coupon tests Yes Damage tolerant development tests Yes Damage tolerance analysis Yes Damage tolerant full-scale component tests Yes Implement IDPP In work at time of cancellation NDE of flight parts Yes Acceptance proof test Yes Post-proof NDE Yes In-service inspection NA (single flight part) * Assessment of MSFC/R. Wingate.

1.3.2.2 Ares I First Stage Frustum. Design work for the Ares I First Stage Frustum began after the release of MSFC-RQMT-3479 in 2006 and thus had those requirements levied on the project via NASA-STD-5019. The MSFC FCB approved the fracture control plan on April 21, 2011, in memo EM20-11-FCB-006.

9 Table 5. Ares I First Stage Frustum mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA Yes DPP. Yes Damage tolerant coupon tests Yes Damage tolerant development tests Yes Damage tolerance analysis Yes Damage tolerant full-scale component tests Yes Implement IDPP In work at time of cancellation NDE of flight parts Yes Acceptance proof test Yes Post-proof NDE Yes In-service inspection NA (single flight part)

* Assessment of MSFC/D. Phillips based on information collected by MSFC/R. Wingate.

1.3.2.3 Space Shuttle External Tank Fwd Gaseous Hydrogen Press Line Fairing. Design work for the Space Shuttle ET Fwd GH2 press line fairing began around 1985. MSFC-RQMT-3479 did not exist yet. The damage tolerance approach for this project would today be considered an alternate approach per MSFC-RQMT-3479.

Table 6. Space Shuttle ET Fwd GH2 Press Line Fairing mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA Unknown IDPP Unknown Damage tolerant coupon tests No Damage tolerant development tests No Damage tolerance analysis No Damage tolerant full-scale component tests No Implement IDPP Unknown NDE of flight parts Yes Acceptance proof test Unknown Post-proof NDE Unknown In-service inspection NA (single flight part)

* Assessment of MSFC/D. Phillips based on information collected by MSFC/R. Wingate.

10 1.3.2.4 Space Shuttle External Tank Intertank Access Door. Design work for the Space Shuttle ET intertank access door began around 1985 with damage tolerance requirements per MMC-ET-SE13-C, “Space Shuttle External Tank Fracture Control Program Requirements and Implementation Document” and MSFC-HDBK-1453, Fracture Control Program Requirements. This was also before MSFC-RQMT-3479, “Fracture Control Requirements for Composite and Bonded Vehicle Payload Structures.” The damage tolerance approach for this project would today be considered an alternate approach per MSFC-RQMT-3479. The damage tolerance approach for the intertank access door was reviewed and approved by the MSFC FCB in October 1991.

Table 7. Space Shuttle ET Intertank Access Door mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA Intent** (potential flaw types defined) IDPP Intent (photos show protective cover) Damage tolerant coupon tests Yes Damage tolerant development tests No Damage tolerance analysis Intent Damage tolerant full-scale component tests No Implement IDPP Intent NDE of flight parts Yes Acceptance proof test No Post-proof NDE No In-service inspection NA (single flight part)

• Assessment of MSFC/R. Wingate. ** Intent means engineering activities were conducted that are considered to partially address the intent of the MSFC‐RQMT‐3479 requirement, though methods used may not be what would be used in a current effort. MSFC‐RQMT‐3479 did not exist at the time of ET composite hardware development, but the principal concepts of composite damage tolerance existed well before MSFC-RQMT‐3479.

1.3.2.5 Space Shuttle External Tank Composite Nose Cone. Design work for the Space Shuttle ET composite nose cone began around 1985 with damage tolerance requirements per MMC-ET-SE13-C, “Space Shuttle External Tank Fracture Control Program Requirements and Implementation Document” and MSFC-HDBK-1453, “Fracture Control Program Requirements.” This project was also pre-MSFC-RQMT-3479. The damage tolerance approach for this project would today be considered an alternate approach per MSFC-RQMT-3479. The damage tolerance approach for the composite nose cone was reviewed and approved in the mid-1990s.

11 Table 8. Space Shuttle ET composite nose cone mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA Intent** (potential flaw types defined) IDPP Unknown Damage tolerant coupon tests Yes Damage tolerant development tests No Damage tolerance analysis Intent Damage tolerant full-scale component tests Yes Implement IDPP Unknown NDE of flight parts Yes Acceptance proof test No Post-proof NDE No In-service inspection NA (single flight part)

* Assessment of MSFC/R. Wingate. ** Intent means engineering activities were conducted that are considered to partially address the intent of the MSFC‐RQMT‐3479 requirement, though methods used may not be what would be used in a current effort MSFC‐RQMT‐3479 did not exist at the time of ET composite hardware development, but the principal concepts of composite damage tolerance existed well before MSFC-RQMT‐3479.

1.4.2.6 Space Shuttle Filament Wound Case. Design work for the Space Shuttle FWC began in the early 1980s (the first flight was to have been in 1986). MSFC-RQMT-3479 was not in existence. Fracture control requirements were required to be specified in a supplier-prepared fracture control plan, and damage tolerance was to be determined by methods approved by Morton Thiokol and MSFC. The damage tolerance approach for this project would today be considered an alternate approach per MSFC-RQMT-3479.

Table 9. Space Shuttle FWC mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA Intent** (potential flaw types defined) IDPP Yes*** Damage tolerant coupon tests Yes Damage tolerant development tests Yes Damage tolerance analysis Yes Damage tolerant full-scale component tests Yes Implement IDPP Yes NDE of flight parts Yes Acceptance proof test Yes Post-proof NDE Yes In-service inspection N/A (single flight part)

* Assessment of MSFC-Auburn/F. Ledbetter ** Intent means that engineering activities were conducted that are considered to partially address the intent of the MSFC‐RQMT‐3479 requirement, though methods used may not be what would be used in a current effort. MSFC‐RQMT‐3479 did not exist at the time of Space Shuttle FWC hardware development, but the principal concepts of composite damage tolerance existed well before MSFC-RQMT‐3479. *** Although not documented in this presentation, it is known that the FWC Program developed a protective cover for each case segment for all transportation operations.

12 1.3.2.7 Hubble Space Telescope Super Lightweight Interchangeable Carrier. The HST SLIC flew in 2009. MSFC-RQMT-3479 was not in existence when the design began; it was required to conform to NASA-STD-5003, “Fracture Control Requirements for Payloads Using the Space Shuttle.” The damage tolerance approach for this project would today be considered an alternate approach per MSFC-RQMT-3479. The damage tolerance approach for the SLIC was approved by an FCB.

Table 10. HST SLIC mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA Intent** (potential flaw types defined) IDPP Yes Damage tolerant coupon tests No Damage tolerant development tests No Damage tolerance analysis No Damage tolerant full-scale component tests Intent*** Implement IDPP Yes NDE of flight parts Yes Acceptance proof test Yes† Post-proof NDE Yes In-service inspection NA (single flight part)

* Assessment of GSFC/K. Segal. ** Intent means that engineering activities were conducted that are considered to partially address the intent of the MSFC‐RQMT‐3479 requirement, though methods used may not be what would be used in a current effort. MSFC‐RQMT‐3479 did not exist at the time of HST SLIC, but the principal concepts of composite damage tolerance existed. *** Full-scale-sized, flight-like component tests performed for allowables; manufacturing defects would be present. † Designed for high margins of safety and acceptance proof test at 1.2 × design limit load on the protofli- ght structure performed to address damage risks.

1.3.2.8 Orion Multi-Purpose Crew Vehicle. Design work for the Orion MPCV began c. 2006. MSFC-RQMT-3479 was levied on the project.

Table 11. Orion MPCV mapping to MSFC-RQMT-3479.*

Assessment Completed? DTA Yes IDPP Yes Damage tolerant coupon tests Yes Damage tolerant development tests Yes Damage tolerance analysis Intent Damage tolerant full-scale component tests Fracture critical proof; structural element only Implement IDPP Yes NDE of flight parts Yes Acceptance proof test Yes Post-proof NDE Yes In-service inspection N/A (single flight part)

* Assessment of GRC-LMC/J. Thesken. 13 2. MATERIALS

The primary goals of the CEUS project materials WBS included (1) materials selection/ procurement and (2) coordination of a laminate equivalency test activity. Manufacturability, co-cure compatibility and, with the prepreg, availability of existing data were factored into the material selection. Availability of an existing materials database enabled equivalency testing as a means to reduce the scope of coupon testing. The prepreg selected for the project was IM7/8552-1, a variant of the IM7/8552 material used to generate test data reported within a database established by the National Center for Advanced Materials Performance (NCAMP). Laminate panels for the CEUS project were fabricated at three NASA Centers to determine equivalency of the material to the database and additionally, equivalence of fiber placement fabrication available at MSFC and LaRC. Coupons from the laminate panels were machined and tested at the National Institute of Aviation Research (NIAR). Statistical analysis methods were used to determine equivalency between the IM7/8552-1 coupons and the NCAMP database for IM7/8552.4 Equivalence of the remotely manufactured composite panels was evaluated by Analysis of Variance (ANOVA).

Along with leveraging the NCAMP database to establish design allowables, a CEUS objective was to evaluate Digimat from e-Xstream engineering/MSC-Software®, a nonlinear multiscale material modeling tool, as a method to provide fiber-reinforced polymer matrix composites for both establishing initial design allowables and enabling an ‘as-built’ digital material for in-service performance predictions.

2.1 Selection and Preexisting Data

Hexcel’s IM7/8552-1 prepreg tape was selected as the STA facesheet material based on its amenability to fiber placement and the available NCAMP materials database. Hexcel’s 8552-1 epoxy resin is a variant of the baseline 8552 resin and was designed for fiber placement. Compared to 8552, the 8552-1 variant demonstrates a lower tack, facilitating movement through the fiber placement head. As data for the 8552 form of the material is available through the NCAMP database,4 the project adopted an accelerated building block approach in the form of an equivalency test matrix to reduce risk related to materials and schedule. The Composite Material Handbook -17 allows equivalency to be demonstrated for design allowables in the case where the differences between the original and new material and/or process are minimals.5

Cytec’s FM300-2M, with 0.06-psf areal weight, was selected as the film adhesive for sandwich panel construction. Cytec’s FM309-1 was initially selected, but the quoted lead time was prohibitive. The FM300-2M film adhesive is a 121 oC cure variant of Cytec’s FM300M adhesive. Previous work had focused on FM300M; however, the 177 oC cure led to co-mingling of the adhesive and the epoxy resin from the facesheet. The 121 oC cure variant was selected to initiate adhesive cure prior to that of the facesheet, and reduce mixing of adhesive and resin during cure. For this project, there was no investigation of the mat-versus-knit carrier, but such a study could have been beneficial, in particular, where foam core may be considered.

14 The 20-day out-time specified by the vendor was a concern with the FM300-2 film adhesive. An out-time study was carried out to address these concerns; the results are outlined in section 2.4.

Aluminum (Al) honeycomb core was procured from Alcore in flat sheets for process development. The core was phosphoric acid, anodized 5052 aluminum honeycomb, perforated, 1 in thick and 1/8 in cell size.

2.2 Prepreg Procurement Specification

The IM7/8552-1 prepreg material was ordered from Hexcel. The 12-in parent tape was fabricated at Hexcel Corp, Salt Lake City, Utah, and slit at Web Industries, Atlanta, Georgia. The slit tape width specifications included 1/4-in-wide tape provided to LaRC and a 1/2 -in-wide tape provided to MSFC.

There was considerable discussion within the project regarding the development of an internal NASA specification for the prepreg material. Ultimately, the schedule did not allow for such development, particularly as a portion of the material was procured early within the project for process development and equivalency test panels. Deviating from the specification of the first procurement was not an option.

2.3 Storage and Handling

Guidelines were established to address risk associated with material storage and handling. Specifically, the document described the requirements and methods for storage and handling of temperature-sensitive materials, including Hexcel’s Hexply 8552-1, carbon fiber-reinforced epoxy prepreg. The scope of the storage and handling document included requirements for inspection on material delivery, usage, and requalification of a material that exceeded the manufacturer specified out-life or shelf life. Throughout the CEUS project, prepreg and film adhesive were stored in a freezer, at or below 0 oF, and out-time was recorded as the material was used. Acceptance tests included the Differential Scanning Calorimetry (DSC) test for extent of cure and the High-Pres- sure Liquid Chromatography test for relative constituent concentrations, as well as physical and mechanical test data provided by Hexcel on delivery. This activity was coordinated with Safety and Mission Assurance (SMA). A full description of the SMA activities within the CEUS project is provided in section 4.

2.4 Film Adhesive Out-Time Study

The CEUS manufacturing schedule estimated 20 days to manufacture each 1/8-in skirt segment; the film adhesive would sit on the tool through most of that manufacturing process. During this out-time period, the physical and/or chemical properties of the material may change as a result of room temperature cure advancement. A study on the room temperature stability of the film adhesive was initiated to mitigate risk related to film adhesive out-time and to ensure that material manufacturability, strength, durability, and other physical properties are not adversely affected by out-time. The film adhesive was aged at room temperature for 30 days, and the thermal

15 response of both the baseline and aged materials was evaluated by DSC and rheology measurements. Flat-wise tensile strength of sandwich panels fabricated from baseline and aged adhesive was also evaluated. Photos of failed coupons are shown in figure 3. Both core and adhesive failure was observed within the set of coupons tested with 20-day-aged adhesive; however, there was no overall drop in flatwise tensile strength. In conclusion, the study found no effect on the degree of cure nor reaction temperature following a 30-day out-time. The rheological cure behavior showed a constant gelation temperature through 30 days of out-time, whereas the vitrification temperature decreased with out-time. The time to gelation and vitrification also decreased with increasing out- time. Flatwise tensile strength showed no significant change with film adhesive out-time; however, failure was primarily observed within the 4.5-pcf core.

Figure 3. Core and adhesive failure within the coupons fabricated from 20-day-aged film adhesive.

2.5 Equivalency Tests

The purpose of the equivalency test matrix was to reduce the scope of coupon testing within the CEUS project and enable application of the NCAMP-generated materials data to the STA design. Panels for equivalency tests were fabricated at three NASA Centers with the goal of dem- onstrating material equivalency and equivalency of the manufacturing method, i.e., fiber placement to hand layup. Panels from each Center were shipped to the NIAR where coupons were machined, conditioned, and tested.

2.5.1 Panel Processing Specification

A processing specification was established to ensure consistency of fabrication and processing methods between remote Centers. The document, available on SharePoint, was based off of an NCAMP provided processing document, thereby ensuring consistency with panels fabricated for database development.6

16 2.5.2 Equivalency Panel Fabrication

Figure 4 details the bagging arrangement used to manufacture equivalency test panels. The cure cycle followed NCAMP processing conditions. This cure profile, identified as ‘baseline/ medium cure cycle (M),’ varied from the vendor-recommended cycle.

Baseline/Medium Cure Cycle (M): Check vacuum bag integrity prior to starting the cure cycle; leak rate shall not exceed 5 in Hg in 5 minutes. All temperatures are part temperatures. Steps (1)–(6) are based on leading thermocouple, except step 5 is based on lagging thermocouple.

(1) Pull vacuum (min 22 in Hg).

(2) Heat at 2o F/min to 355 ± 10 oF, and ramp autoclave pressure to 100 psig.

(3) Before temperature reaches 140 oF and when autoclave pressure is 20 ± 10 psig, vent vacuum bag to atmosphere.

(4) From 325 oF to 355 ± 10 oF, a minimum heat up rate of 0.3 oF/min is acceptable.

(5) Hold 355 ± 10 oF for 120 +60°/–0 min.

(6) Cooldown rates from cure temperature to 150 oF shall be no more than 10 oF/min.

(7) Release autoclave pressure when lagging thermocouple is below 150 oF or minimum of 1 hr into cooldown, whichever occurs sooner.

(8) Remove from autoclave when autoclave temperature is less than 120 oF.

Vacuum Vacuum Bag Breather

Caul Tape Seal Nonporous FEP** Release Fabric*

2Breather String Part

Silicone Release Fabric* Sealant Nonporous FEP Bottom Tool * Release fabric is to be used on 0°, unidirectional tension (longitudinal tension, LT) panels only. ** The breather string must be in the edge of the part, not laid on the top of the panel, and must extend out past the sealF4_1624 to touch the breather pad material as shown. Figure 4. Recommended bagging sequence.

17 LaRC and MSFC each fabricated three equivalency test panels by fiber placement, where panel dimensions were determined by the dimensions and quantity of coupons required for mechanical tests. Due to the size of the GRC autoclave, smaller panels were fabricated and a total of 16 panels were made to meet the coupon requirements. The key difference between these three sets of panels was the use of 1/4-in-wide slit tape at LaRC, 1/2-in-wide slit tape for fiber placement at MSFC, and 12-in-wide unidirectional prepreg for hand layup at GRC.

Panel Fabrication, LaRC: The ISAAC system7 (fig. 5) at LaRC was used to fabricate three panels (fig. 6) from 1/4-in-wide IM7G/8552-1 graphite/epoxy prepreg slit tape.8 The panels included a unidirectional 6-ply panel (14 in×26 in), a quasi-isotropic 16-ply panel (24×24 in2), and a quasi-isotropic 24-ply panel (24 in×24 in).

Figure 5. LaRC ISAAC system.

18 Figure 6. Equivalency panel fabrication at LaRC.

The panel layups were first programmed using the CGTech VERICUT® for Composites Programming (VCP) software, tested virtually prior to running on the ISAAC hardware using the CGTech VERICUT for Composites Simulation (VCS) software (fig. 7), and tested virtually (fig. 8) prior to running on the ISAAC hardware using the CGTech VCS. After layup on ISAAC, the panels were then cured at LaRC using the cure cycle specified by the CEUS processing document.

Figure 7. Panel layups tested on CGtech VCP.

19 Figure 8. CGTech VERICUT tested virtually.

Panel Fabrication, MSFC: Details of MSFC panel fabrication will be provided in the manufacturing section.

Panel Fabrication, GRC: Panels were hand laid from 12-in-wide parent tape and followed the CEUS processing document. Panels were a maximum of 12×12 in, requiring 16 panels to be fabricated for equivalency testing.

2.5.2.1 Panel Quality. Panel quality was characterized at NIAR by ultrasonic C-scan to ensure acceptable consolidation prior to test. Representative images from panels fabricated at GRC, LaRC, and MSFC are shown, respectively, in figure 9. These images represent the [45/0/–45/90]2s ply configuration.

(a) (b) (c)

F9_1624 Figure 9. C-scan images of the 16-ply panels fabricated for (a) UNT, (b) UNC, and (c) OHT tests.

20 Coupons from each panel were sectioned for optical microscopy and acid digestion. Representative photomicrographs in figure 10 support NDE indicating low void content throughout the panels. Acid digestion showed less than 1% void content in panels evaluated.

(a) (b)

(e) (f)

Figure 10. Photomicrograph of GRC panels: (a) 24 ply and (b) 36 ply, LaRCF10_1624 panels: (c) 24 ply and (d) 36 ply, and MSFC panels: (e) 24 ply and (f) 36 ply.

21 2.5.3 Test Matrix and Data

The equivalency test matrix outlined in table 12 reflects a portion of the data available within the NCAMP database. The identified coupon tests were chosen to provide confidence in the material and manufacturing method. Ply configurations matched those used to generate the NCAMP database. Coupons were machined by wet diamond saw to be oversized, then ground to meet American Society for Testing and Materials (ASTM) specifications. Details on coupon size are also outlined in table 12. Test conditions matched those used in the NCAMP database, where the cold temperature dry condition was defined as –65o F, room temperature dry (RTD) condition was 70 o F, and the elevated temperature wet (ETW) condition was 250 oF.

Table 12. Test matrix.

Specimen Specimens Panel Thickness Size (in) Test Nom. per Spare per No. of Thick Test Description Test Standard Test Type Condition Center Center Plies (in) 0 Dir. 90 Dir. Solid Panels Lamina-tension, 0-deg ASTM D3039 T CTD 6 1 6 0.04 10 1 direction RTD 6 1 ETW 6 1 Laminate tension (Q/I) ASTM D3039 UNT CTD 6 1 16 0.12 10 1 RTD 6 1 ETW 6 1 Laminate compression ASTM D6641 UNC CTD 6 1 16 0.12 5.5 0.5 (Q/I) RTD 6 1 ETW 6 1 Short beam shear strength ASTM D2344 S CTD 6 1 24 0.17 1.5 0.5 RTD 6 1 ETW 6 1 Open-hole compression ASTM D6484 OHC CTD 6 1 24 0.17 12 1.5 RTD 6 1 ETW 6 1 Open-hole tension ASTM D5766 OHT CTD 6 1 16 0.12 12 1.5 RTD 6 1 ETW 6 1 Compression after impact ASTM D7137 CAI CTD 6 1 24 0.14 6 4 RTD 6 1 ETW 6 1

22 Mechanical test data were generated for coupons fabricated at each Center and preliminary statistical analysis determined a ‘per-Center’ equivalency to the NCAMP database. The coupons prepared at separate locations could not be grouped for equivalency analysis because doing so would imply equivalency of those panels. Panels made at separate locations could not be assumed to be equivalent; a component of this effort was to demonstrate equivalency of the processing methods. While each center routinely passed the equivalency metric, that pass came with the caveat of ‘insufficient data.’ Per CMH-17 guidelines, eight coupons are required to determine equivalency, and each Center individually offered only four to six coupons per test.

ANOVA was performed on the test data to determine statistical equivalence of coupons from separate Centers, therefore allowing those coupons to be grouped in the analysis for equivalency. ANOVA is a well-known statistical method used to determine whether independently fabricated panels can be considered equivalent, although this level of equivalency is, in general, very difficult to establish. Equivalency between even two of the Centers would provide a sufficient coupon count to remove the ‘insufficient data’ descriptor. ANOVA allowed data pooling in most cases. Where allowed, these data are presented in tables 13–18 as ‘Combined Data.’

The mechanical test data generated by NIAR is tabulated in table 13, with the Pass/Fail col- umn indicative of the equivalency metric. For statistical analysis for equivalency testing of composite materials, alpha is set at 0.05, which corresponds to a confidence level of 95%. This means that if the null is rejected and the two materials are not equivalent with respect to a particular test, the probability that this is a correct decision is no less than 95%.

The NIAR report utilized a modified coefficient of variation (CV); in accordance with section 8.4.4 of CMH-17 Rev. G. It is a method of adjusting the original basis values downward in anticipation of the expected additional variation. Composite materials are expected to have a CV of at least 6%. When the CV is less than 8%, a modification is made that adjusts the CV upwards. Several datasets in the CEUS equivalency matrix passed the equivalency standard under the Modi- fied CV condition. Complete details on this method are provided in the NIAR report.

The mean value of each dataset is tabulated in tables 13–18, with the coefficient of variation noted parenthetically. The following items are important to note:

(1) Different materials were used. Hexcel expects a reduction in compressive strength of the 8552-1 variant compared to the baseline 8552 matrix resin. Internal Hexcel testing has confirmed this to be the case; however, modification of the test method reduced the gap in measured compressive strength between the two materials.

(2) The statistical analysis was performed with cured ply thickness (CPT) normalized to 0.0072 in. The average CPT of NASA-generated panels was 0.0070 in.

(3) All ‘passes’ are with the ‘Insufficient Data’ caveat, and a ‘pass’ with Modified CV was considered a ‘pass’.

23 Table 13. Lamina tensile strength and modulus data.

CTD RTD ETW Panel Raw Data Normalized Raw Data Normalized Raw Data Normalized [0]6 (CV) Data Pass/Fail (CV) Data Pass/Fail (CV) Data Pass/Fail NCAMP Database Strength (ksi) 353.7 357.4 (6) – 371.1 362.7 (6.2) – 328.0 33.5 (11.7) – Modulus (Msi) 22.3 22.6 (1.7) – 23.5 23.0 – 23.8 24.0 – COMBINED DATA Strength (ksi) – 363.3 (3.0) Pass – 389.1 (3.1) Pass – 360.0 (3.5) Pass Modulus (Msi) – 22.3 (1.6) Pass – 22.4 (1.6) Pass – 23.1 (1.5) Fail HXL-H12-GRC Strength (ksi) 404.5 (3.7) 363.7 (3.5) Pass 427.1 (1.4) 388.3 (1.7) Pass 394.3 (3.5) 356.4 (3.2) Pass Modulus (Msi) 24.7 (0.7) 22.2 (0.6) Pass 24.8 (0.8) 22.5 (2.2) Pass 25.7 (1.3) 23.2 (1.7) Pass HXL-H12-LaRC Strength (ksi) 387.0 (3.4) 368.2 (3.2) Pass 412.1 (3.7) 381.7(14.2) Pass 380.8 (3.8) 358.1 (3.7) Pass Modulus (Msi) 23.3 (1.0) 22.1 (0.9) Pass 24.1 (2.4) 22.4 (2.4) Pass 24.6 (1.8) 23.2 (1.6) Pass HXL-H12-MSFC Strength (ksi) 371.5 (1.8) 358.0 (1.9) Pass 416.8 (3.7) 397.1 (2.7) Pass 380.8 (3.5) 366.5 (3.5) Pass Modulus (Msi) 23.0 (1.1) 22.1 (1.0) Pass 23.7 (1.6) 22.6 (0.3) Pass 23.7 (1.0) 22.8 (0.9) Pass

Table 14. Unnotched tensile strength and modulus data.

CTD RTD ETW Panel Raw Data Normalized Raw Data Normalized Raw Data Normalized [45/0/–45/90]2s (CV) Data Pass/Fail (CV) Data Pass/Fail (CV) Data Pass/Fail NCAMP Database Strength (ksi) – 97.2 (6.1) – – 103.4 (7.1) – – 110.5 (5.8) – Modulus (Msi) – 8.1 – – 8.2 – – 7.6 – COMBINED DATA Strength (ksi) – 108.8 (2.3) Pass – 106.9 (3.9) Pass – Could not NA pool data Modulus (Msi) – Could not NA – 8.1 (3.6) Fail – 8.0 (1.5) Pass pool data HXL-H12-GRC Strength (ksi) 116.8 (3.7) 109.9 (3.0) Pass 113.7 (1.7) 105.3 (6.7) Pass 118.4 (2.9) 108.3 (3.0) Pass Modulus (Msi) 8.7 (2.2) 8.2 (1.4) Pass 8.6 (3.7) 7.9 (5.7) Pass 8.7 (1.8) 8.0 (1.0) Pass HXL-H12-LaRC Strength (ksi) 110.1 (1.8) 107.1 (1.4) Pass 111.0 (1.6) 108.0 (1.4) Pass 119.3 (1.9) 116.1 (2.4) Pass Modulus (Msi) 8.7 (1.6) 8.4 (1.3) Pass 8.4 (1.8) 8.2 (2.0) Pass 8.2 (1.4) 8.0 (1.8) Pass HXL-H12-MSFC Strength (ksi) 112.8 (1.6) 109.5 (1.6) Pass 110.5 (1.4) 107.4 (1.5) Pass 114. (2.6) 110.9 (2.5) Pass

Modulus (Msi) 8.6 (0.5) 8.3 (0.5) Pass 8.3 (1.4) 8.1 (1.2) Pass 8.1 (1.8) 7.9 (1.6) Pass

24 Table 15. Unnotched compression strength and modulus data.

CTD RTD ETW Panel Raw Data Normalized Raw Data Normalized Pass/ Raw Data Normalized [45/0/–45/90]2s (CV) Data Pass/Fail (CV) Data Fail (CV) Data Pass/Fail NCAMP Database Strength (ksi) – – – – 87.1 (9.3) – – 57.7 (11.0) – Modulus (Msi) – – – – 7.6 (4.8) – – 7.1 (1.8) – COMBINED DATA Strength (ksi) – – – – 92.4 (5.1) Pass – 57.7 (8.6) Pass Could not Modulus (Msi) – – – – 7.7 (1.8) Pass – pool data NA HXL-H12-GRC Strength (ksi) 112.8 (4.5) – _ 95.1 (4.3) 94.5 (7.8) Pass 59.6 (12.6) 60.5 (12.7) Pass Fail Modulus (Msi) 7.9 (1.9) – _ 8.0 (0.8) 8.0 (4.5) Pass 7.7 (1.3) 7.8 (1.6) (2.9%) HXL-H12-LaRC Strength (ksi) 119.8 (3.7) – _ 94.1 (2.2) 92.2 (2.3) Pass 57.9 (7.5) 56.4 (7.5) Pass Modulus (Msi) 7.9 (1.2) – _ 7.8 (1.3) 7.6 (1.3) Pass 7.6 (0.9) 7.4 (0.8) Pass HXL-H12-MSFC Strength (ksi) 114.3 (5.3) – _ 97.7 (3.3) 95.0 (3.3) Pass 62.1 (2.7) 60.3 (2.8) Pass Modulus (Msi) 8.1 (0.7) – _ 7.9 (0.4) 7.7 (0.5) Pass 7.7 (1.2) 7.5 (1.2) Pass

Note: The fail in the compression modulus is due to a measured value that is greater than the modulus reported in the database.

Table 16. Laminate short beam shear data.

CTD RTD ETW Panel Raw Data Normalized Raw Data Normalized Raw Data Normalized [45/0/–45/90]3s (CV) Data Pass/Fail (CV) Data Pass/Fail (CV) Data Pass/Fail NCAMP Database Strength (ksi) – – – NA 12.13 (6.9) – – 6.99 (3.7) – COMBINED DATA Strength (ksi) – – – – 12.3 (3.7) Pass – 6.8 (3.6) Fail HXL-H12-GRC Strength (ksi) 12.89 (2.9) – – 12.45 (3.2) 12.4 (3.2) Pass 6.8 (4.0) 6.8 (4.4) Pass HXL-H12-LaRC Strength (ksi) 11.31 (7.8) – – 12.03 (3.9) 12.0 (3.9) Pass 7.0 (2.3) 7.0 (2.3) Pass HXL-H12-MSFC Fail Strength (ksi) 13.55 (2.2) – – 12.4 (3.8) 12.4 (3.8) Pass 6.7 (3.1) 6.7 (3.2) (0.1%)

25 Table 17. Open-hole tension data.

CTD RTD ETW Panel Raw Data Normalized Raw Data Normalized Raw Data Normalized [45/0/–45/90]2s (CV) Data Pass/Fail (CV) Data Pass/Fail (CV) Data Pass/Fail NCAMP Database

Strength (ksi) 57.8 (4.2) – 59.0 (4.0) – 67.0 (4.3) – COMBINED DATA Strength (ksi) Could not 63.6 (2.6) Pass pool data NA 69.2 (4.4) Pass HXL-H12-GRC

Strength (ksi) 64.6 (2.3) Pass 66.0 (2.8) Pass 69.2 (6.9) Pass HXL-H12-LaRC Strength (ksi) 63.6 (2.8) Pass 62.4 (1.8) Pass 69.0 (2.4) Pass HXL-H12-MSFC Strength (ksi) 62.6 (2.2) Pass 64.1 (2.6) Pass 69.4 (3.4) Pass

Table 18. Open-hole compression data.

CTD RTD RTD ETW ETW Panel Raw Data Normalized Raw Data Normalized Pass/ Raw Data Normalized [45/0/–45/90]3s (CV) Data Pass/Fail (CV) Data Fail (CV) Data Pass/Fail NCAMP Database – – – – – – – – – Strength (ksi) – – – – 49.1 (3.7) NA – 35.5 (4.1) – COMBINED DATA Strength (ksi) – – – – 46.9 (2.8) Fail – 32.5 (2.9) Fail HXL-H12-GRC Strength (ksi) – – – – 45.8 (2.1) Fail – 31.8 (1.4) Fail (1.1%) (5.1%) HXL-H12-LaRC Strength (ksi) – – – – 47.1 (2.9) Pass – 32.2 (2.4) Fail (4.8%) HXL-H12-MSFC Strength (ksi) – – – 47.5 (3.7) Pass – 33.2 (2.7) Fail (1.9%)

NIAR characterized a failure as mild if the percent fail is ≤4% for modulus and ≤5% for strength. The majority of failures observed in this study would be considered mild. The Glenn open-hole compression (OHC)/ETW coupons had a percentage failure of 5.1%, which is characterized as a mild to moderate failure. Although generally mild, the failures in the equivalency test matrix are addressed below.

26 The tests that failed were the (1) unnotched compression modulus at elevated temperature, (2) short beam shear (SBS) at elevated temperature, and (3) open-hole compression. The failure in unnotched compression was due to a modulus value that exceeded the database value. A higher value is not an issue.

The SBS and OHC failures are likely related to the variation of the matrix resin, as predicted by the vendor.

The NIAR statistical report, which includes the ANOVA analysis, is presented as appendix B. The report includes data for GRC-generated coupons labeled as (1) slow ramp and (2) low vacuum. The low vacuum coupons are mislabeled and should read ‘low pressure.’ These coupons were tested 1 to represent realistic manufacturing conditions of the STA. The mass of the STA /8 arc segment would likely slow the part ramp temperature to less than the 2 oF/min used to generate equivalency data. The ‘slow-ramp’ coupons were generated using a 0.5 oF/min ramp rate. In addition, the pressure on the sandwich panel facesheet will be in the range of 45 psi, as opposed to the 100 psi used to manufacture the equivalency panels, and this condition is represented by the ‘low vacuum (pressure)’ coupons. As these coupons were outside the scope of the equivalency matrix, they were not included in the above tables.

2.6 Safety and Mission Assurance

The CEUS and later Composites for the Universal Stage Adapter 2 (CUSA2), and later the SMA effort, was an initiative including multiple NASA Centers. MSFC led a collaborative bimonthly discussion between LaRC, GRC, and Goddard Space Flight Center (GSFC). The col- laboration provided a synergistic opportunity for the Centers to compare and choose the best path between respective Center quality plans and best practices.

Project deliverables, e.g., SMA and test plans, were generated in support of the CEUS- CUSA2 project plan to advance the technology readiness level (TRL) (see app. C) of composites for launch vehicles from TRL 5 to TRL 6, initially planned to complete in time for the EUS preliminary design review (PDR) and potential infusion onto the SLS. Additionally, a qualification path leading to potential human-rating certification requirements for composites was proposed as a key deliver- able, but the project was cancelled before its completion. In fulfillment of the CEUS-CUSA2 SMA requirements before its cancellation, input was provided in the “Processes for Storage and Handling of Prepreg Including Guidelines for Receiving Inspection and Re-Qualification of Carbon Fiber/ Epoxy Prepreg Materials” led by the GRC CEUS-CUSA2 material lead. GRC SMA developed a ‘quality engineering and assurance checklist’ for the coupon testing scheduled to be performed. Either the SMA lead or a delegate, reviewed the CEUS-CUSA2 requirements. Receiving inspections were performed at delivery.

27 The MSFC Science and Mission Systems Assurance Branch (QD22) prepared both the SMA and Mishap plans to define the Industrial Safety and Quality Plan for the CEUS-CUSA2 project. The successful implementation of the SMA plan required a coordinated effort from all members of the CEUS-CUSA2 team involved in the proposed procurement, testing, handling, and storage of the CEUS-CUSA2 coupons, test panels, and static test articles. CEUS-CUSA2 test articles were developmental/nonflight; however, at one point, various configurations were proposed ranging from a USA2 prototype static test article to single test panels. The associated costs of these proposed configurations ranged from approximately $20 million to $50 million. SMA was prepared to meet the needs of the CEUS-CUSA2 project whichever path was chosen. The final configuration resided along the less expensive test article path.

The SMA lead coordinated the development of the SMA plan across the project and other Centers with the CEUS PM and CE. The SMA lead or quality assurance delegate(s) representing QD22 at each Center were responsible for specific duties and sharing applicable quality records on the CEUS-CUSA2 share site until the date of its cancellation. The duties were as follows:

• Witness testing activities. • Verify that testing procedures were being followed. • Verify test equipment and instrumentation calibration was up-to-date. • Assure that necessary data were recorded and reported. • Approve fabrication requests and manufacturing documents. • Review calibration records and test procedure for testing performed through contract at the NIAR.

2.7 Computational/Model Based Materials

Composite laminate property allowable determination is a core effort for any composite structure development. Test programs are executed in a number of ways, including full test programs based on Composite Material Handbook-17 (CMH-17), with high costs and long execution time. Some programs opt to reuse past databases with no testing, such as when the prime contractor uses in-house, established databases. This is good when a plethora of data have been collected, and the same material and processes are used. Other programs might leverage published data with targeted testing to validate program specific design variants; the CEUS equivalency program is an example. All approaches have a common goal––establishing design allowables that can be used with confidence to predict the service life of composite structures. All of these methods rely on experimental approaches. New computing power and simulation capabilities are enabling finite element (FE) modeling down at the constituent levels, a new world into determining composites material allowability. This enables constituent material modeling to evaluate microscale material effects on the macroscale material response. This digital approach provides the opportunity to capture constituent material variability, process variables, and test variables––traditionally captured through testing multiple batches and many replicates––without all the testing.

28 2.7.1 Digimat (for CEUS)—Introduction

Digimat is a nonlinear, multiscale, material and structural modeling platform that allows for describing micro- and macro-level composite behavior and for bridging the gap between manufacturing and performance. A 1-year lease of the tool allows for evaluating the software capabilities that are in line with CEUS scopes during 2016. Digimat offers several tools and solution modules for several applications (fig. 11). The following is a brief description of each module:

Figure 11. Digimat platform and available modulus at GSFC.

• Digimat-MF: Computes the macroscopic performance of composite materials from their per- phase properties and microstructure definition. • Digimat-FE: Microscopic level FE approach to obtain an in-depth view into the composite material by the direct investigation of representative volume elements (RVEs). • Digimat-VA: Computes, instead of tests, the behavior of composite coupons to screen, select and compute allowables of composite materials. • Digimat-CAE: Interface to finite element analysis (FEA) packages for structural analysis level.

As mentioned, at the CEUS, an objective was defined to demonstrate and validate the software capabilities. Some of the potential benefits of adapting this tool, after verification, are as follows:

• The capability of computing (instead of extensive testing) statistical basis material allowables: – Compare to CUSA material (equivalency) test data (established for analysis). – How does a Digimat computed B-Basis allowables (using 8552-1 limited test data) compared with the published 8552 B-Basis allowables?

29 • Being able to understand material propertie’s sensitivity to manufacturing processes, such as void content and fiber volume fraction: – How well the imperfection-affected properties computed by Digimat compare to the test data. (See demo 1.) • Executing accelerated (instead of traditional) building block with virtual tests and reduce test matrices: – Using unnotched test data to statistically compute open-hole allowables. • The possibility of efficient materials development across programs (cost and schedule), e.g., design the architecture of 3D woven composites for joint applications. (See demo 2.)

Demo Example 1: Using Digimat-FE, it is possible to create/design an RVE of the composite material of interest and compute the composite properties based on its constituent’s properties. Figure 12 illustrates a simple process of creating an RVE, for a unitape material, and extracting the material in-plane and out-of-plane stiffness responses (E11, E22, and E33) under an applied load.

Microstructure Definition/ FE Mesh RVE Creation

Submitted to MSC Marc

FEA Results  5 Material Property Extraction 2.5 10 Comp 11 of Stress  – 89,5815 11 11  – 2 22 22 77,2121  – 33 33 64,8427 1.5 52,4733 40,1038 1 27,7344 Stress (psi) 15,3650 0.5 29,9563 –93,737.8 0 0 0.002 0.004 0.006 0.008 0.01 Strain

Figure 12. Design and material property computation of a unitapeF12_1624 material.

Demo Example 2: Similar to example 1, a more complicated architecture can also be designed, and properties in different directions can be computed. Figures 13 and 14 show the design of a 3D woven RVE and FEA results for material property extractions, respectively. This is of interest to composite joint technology development activity where a preform joint material can be designed (i.e., going through a trade study) to meet predefined requirements, prior to procuring

30 materials and performing any testing. In this particular application, the design-adjustable parameters include: individual phase material properties, different materials for different phases, and fabric characteristics (e.g., number of warp/weft yarns, number of layers, weave steps, warp/weft/linker yarn, warp/weft yarn count, etc.).

FE Mesh

Representative Volume Element (RVE)

X Matrix phase not shown

Figure 13. A 3D woven RVE creation and corresponding FE mesh. F13_1624

×104 4 Comp 11 of Stress Data Y Direction 120,527 3.5 Data Direction 105,331 3 90,135.4 y = 6.8 msi and =1.2 msi 74,939.5 2.5 y – 59,743.7 2 σy ∈ (psi)

44,547.7 σ 1.5 29,351.8 14,155.9 1 σz ∈– z –1,039.97 0.5 Y 0 0 1 2 3 4 5 X ε ×10–3

Figure 14. FEA results and property extraction for a 3D woven RVE. F14_1624

31 2.8 Digimat Test Simulation

Digimat-VA makes it possible to reduce material property characterization test programs by taking advantage of virtual testing and virtual allowable computation. Instead of executing an extensive test program, a selected number of experiments can be performed, and results can be used to compute for the rest. Virtual test matrices can be created, and execution (through FE analysis) of the test matrix provide the selected material virtual allowables. The following general steps are needed to create and execute a virtual test program:

• Test matrix definition includes defining material, layups, type of test, and environmental conditions (fig. 15).

Figure 15. Test matrix definition.

32 • Test variability definition: includes a set of adjustable parameters that accounts for probable variabilities such as constituent variability, process variability, and testing variability (fig. 16).

‘Variability’ Incorporation to Digimat Analysis for Virtual Testing

Figure 16. Test variability definition.

33 • Simulation and allowable computation: Based on the selected variabilities and type of ASTM test, each coupon is analyzed, and a family of allowables will be generated to calculate A- and B-basis allowable, etc. Figure 17 shows the selected variabilities and results for 18 virtual unnotched tension tests, and calculated allowable for IM7-8552 QI [45/–45/90/0]s laminate.

1.4e5

1.2e5

DAMAGE EVOLUTION 1.0e5 Instantaneous VARIABILITY MODEL 0.8e5 ACTIVE PARAMETER VALUE UNITS 0.6e5 Stress (psi) Constituent variability 0.4e5 Matrix Tensile Young modulus Cov 5 % Matrix Compressive Young modulus Cov 3 % 0.2e5 Matrix Tensile Strength Cov 4 % 0 0 0.005 0.010 0.015 Matrix Compressive Strength Cov 2 % Strain

Fiber Tensile Axial Young Modulus Cov 1.5 % Raw data Statistic Value Unit 1.6e5 Mean Fiver Compressive Axial Young Modulus Cov 3 % B-basis Samples 18 Fiber Tensile Strength Cov 6 % A-basis Mean 1.1726E+05 psi 1.4e5 B-basis 93675 psi Fiber Compressive Strength Cov 2.5 % A-basis 76915 psi Process variability 1.2e5 Fiber Volume Fraction Cov 3 % Refresh Testing variability 1.0e5 Fiber Alignment std Deviation 5 %

Strength (psi) 0.8e5

0.6e5

0.4e5

0.2e5

0 0 5 10 15 20 Samples

Figure 17. Simulation and allowable computation. F17_1624

2.8.1 Computational Materials and CEUS Lifecycle

Materials can change over a lifetime of service. The allowables derived from testing during the program inception, as well as derived from ASTM flat panel specimens in the small scale, do not necessarily represent the final part. How the part was made and what the environmental exposures were during storage, testing, and loading can all influence properties. The CEUS computational material plan was to validate the software and collect relevant data during the development lifecycle for an ‘as-built’ structure for evaluation. The as-built data were to include constituent material properties variations in modulus and strength, fiber volumes, and void content of individually built parts. This was to include tool-side and bag-side facesheets for each panel.

34 After manufacturing, environmental exposure would be captured for material updates. These data points were to provide the basis to adjust composite facesheet material properties in Digimat. The updated facesheet material properties applied to a given panel would be input to the structure model for as-built pretest analyses performance predictions. The as-built structure performance model would be used in conjunction with the initial analytical models with pristine material property to evaluate the structure test performance for comparison.

2.8.2 Verification Approach

In order to evaluate the fidelity of the adapted Digimat software, a verification plan, including a test program, was proposed and adopted. Figure 18 shows a complete verification plan with the following three linked phases: (1) Fiber and matrix testing with ‘lamina’-level property predictions and verifications, (2) lamina-level testing with laminate-level property predictions and verifications, and (3) laminate level testing and verifications. Test data generated within this project are presented in appendices A.1, A.2, and A.3.

Limited Fiber and Matrix Testing

RTD CTD ETW Digimat Test Data Test

Lamina Level

Lamina Level Testing Lamina Level (Testing) Pane 1: Min Porosity Digimat Pane 2 & 3: %X and %Y (Cure 1) Porosity (Cure II, III)

RTD CTD ETW FV Test Results RTD CTD ETW Digimat

Laminate Level Testing (Panel I, II, III)

RTD CTD ETW

Figure 18. Digimat software verification approach. F18_1624

35 The plan includes the verification in three different standard environmental conditions— RTD, CTD, and ETW. Characterizing constituents (i.e., fiber and matrix), separately allows for creation (Digimat modeling) of unitape lamina while parameters such as fiber volume (FV), void content, etc., can be accounted for and considered in subsequent analysis. Figure 19(a) shows an RVE with 60% fiber volume fraction, while figure 19(b) illustrates the same RVE with addition of 7% distributed porosity. The idea is to explore whether, for instance, using an input test data of a lamina/laminate with minimum level of porosity (fig. 19(a) to Digimat along with characterized porosity level of another lamina/laminate (fig. 19(b), for instance) can be accurately predicted/ computed.

(a) (b)

Figure 19. RVE: (a) With 60% FV and (b) with 60% FV and 7% void content.

Upon computation of properties of these RVEs, the results can be compared with lamina level test data for verification purposes. Prior to conducting the tests, coupons (or the parent panel) will undergo NDE to characterize the level of porosity and other imperfections.

The verification approach, as shown in figure 18, also scales up the validation approach to lamina and laminate levels where the test data of lower size scale will be used as an input for computing/ predating properties of the next higher size scale constituent. The higher size scale test data will then determine the fidelity of computation/prediction. The proposed complete verification plan entertains the above process using lamina and laminate IM7-8552-1 panels of three different porosity (cure) levels.

36 3. DESIGN

3.1 Composites for Exploration Upper Stage Skirt Design

The initial task for this project was to design, analyze, fabricate, and test two large-scale articles representative of the skirt structures for the EUS, the second stage of the SLS. These skirt structures are now referred to as adapters for clarity in differentiating the structures between the SLS Core Stage and the EUS. Initial concepts for the stage, as shown in figure 20, had a forward skirt approximately 10 ft tall, while the aft skirt was approximately 30 in tall.

Figure 20. Early concept of a cryogenic Upper Stage for the SLS.

37 The launch vehicle skirt application poses an interesting challenge for composite structure. Composite structures used on launch vehicles typically utilize metallic end rings to join to the adjacent component. This may be due to considerations of assembly/disassembly, tolerance/alignment, or stiffness. Because of shrinkage at the cryogenic tank interface, the joint was examined as to efficient methods to accommodate the mismatch. The design baseline was to use a mechanically- fastened aluminum flange that extended until the ambient purge temperatures were reached. Another innovative method was to look at the possibility of tow steering to tailor the effective coefficient of thermal expansion for the structure to more closely align with the hoop contraction witnessed by the LH2 tank.

The aft skirt, in addition to having the same thermal considerations, also had the challenge of interfacing with the structure used to provide the structural interface between the suspended LOX tank and the vehicle OML. This structure delivered out-of-plane loading at locations in the acreage area of the skirt. This would have required significant local reinforcement. Attaching the intertank, or mid-body, type of structure to the aft flange may have been an option to alleviate the imparting of loads in the skirt acreage.

Prior to the SRR for the CEUS, a systems design review was conducted for the EUS. This brought about a major change to the hardware being targeted for composite implementation. The forward and aft skirts were brought to a uniform height of 70 in. While this improved the application for the aft skirt, the loss of 50 in of height for the forward skirt meant that the metallic interface to the tank was a greater percentage of the overall mass of the structure. Due to the integrated SLS vehicle baselining this change, these criteria were adopted for the CEUS activity. This simplified the analysis activity that was being conducted in support of the wall construction trade outlined in section 3.2. Essentially, the higher loads for the aft skirt could then be used as the driving design loads for the trade.

3.2 Wall Construction Trade Study

The wall construction trade study was conducted to assess whether the architecture would be optimal for the EUS skirts. A key criteria, assuming that any structure would be required to satisfy the requirements for structural qualification and material compatibility, is to minimize mass. Other factors must also be considered including cost and schedule to develop. The CEUS project thus undertook a trade with a planned completion date shortly after the SRR was to be conducted. Figure 21 is a slide from the summary charts for the trade indicating the criteria used to assess the options. The analysis for the wall construction trade is detailed in section 4, Joints and Analysis.

38 Figure 21. Composites for Upper Exploration Stage CEUS-TRADE-001.

Considered in the trade were sandwich-type wall constructions, including both honeycomb and foam-reinforced sandwich, as well as stringer-stiffened configurations. The composite material selected was a commercially available, toughened carbon/epoxy. The properties for the composite were extracted from existing databases and assumed an autoclave curing process. Schedule and cost were judged relative to the CEUS project constraints and possibly more limited than an architecture trade for a structure with a one-to-one payload ratio. Cost and schedule thus eliminated all skin-stringer variants for the STA for the CEUS project, due to the process development required for the manual fabrication process.

A concern regarding agency polymer matrix composite development efforts is that each task may not fully utilize the knowledge gained on prior NASA-funded composite projects. This may be more of an artifact of attempting to satisfy aggressive mass- and cost-savings goals for a particular structure, as opposed to heavy-weighing of commonality (e.g., wall construction, material, fabrication techniques, etc.) for a launch vehicle due to cost and schedule concerns. It is often contrasted to metallic hardware design, where it is assumed an alloy and construction (isogrid, orthogrid or hat-stiffened structure) are selected early in the program stages, leading to structural sizing and integrated design activities within months of design authorization. This may be true for applications where mass optimization is not required, but recent experience shows that for metallic structures, alloys and construction are still traded up to milestone reviews, primarily the PDR.

39 Because one of the challenges for composite structures was an efficient design process, one of the goals of this project was to utilize the previous studies to streamline the decision on what construction to utilize for the wall of the structure.

As stated, the trade space for the STA was reduced to selection of core material for a sandwich configuration. A Figure of Merit trade study was discussed, but the decision was made to use an alternate trade form utilized on the early SLS trades which was an advantages/disadvantages trade. In this trade form, the metrics that can be calculated or estimated are reported and then items such as manufacturability, inspectability, or integration of late hardware changes were discussed without assigning a comparative numeric value.

Honeycomb core and foam core structures are both utilized in the aerospace industry, thus indicating that both construction configurations have been manufactured and inspected. For manufacturability, while the two cores are utilized by industry, foam core was preferred over aluminum honeycomb. The rationale is listed in table 19.

Table 19. OHT data.

Foam Core Core used on Delta IV for most all 5-m dry structure. Extensively used by transportation, energy, medical, and architectural industry for lightweight stiffness driver structures. Fabrication requires relatively simple tooling. May require additional step to dry foam before lamination process and seal exposed core immediately. This construction method is the baseline for the established CEUS budget and schedule. Al Honeycomb Extensively used by industry for launch vehicle dry structure fabrication. Lower density core is more prone to damage from handling. Dimpling and ply waviness will occur in piles adjacent to core unless skins are precured. Specific core splice methods are part specific, vendor proprietary, and will require process development to establish a reliable/repeatable core splice application method. Selecting this construction method will exceed the CEUS established schedule.

As listed, any potential advantages of heat leak for one configuration over another was deemed an important criterion for wall-construction selection. The mass savings realized by utilizing composites could thus be enhanced by a reduction in boil-off of hydrogen for on-orbit applications. The Thermal Analysis & Control Branch at MSFC utilized Thermal Desktop to perform a heat leak comparison. The analysis indicated that heat leak during on-pad and ascent may exceed the metallic baseline if a surface treatment was not utilized. On-orbit calculations did indicate potential benefit for composite structures, but at the fidelity of the trade, it was recommended that the foam and aluminum core numbers were within a range that was close enough as to state that heat leak should not be used as a discriminator for core selection.

Damage susceptibility and repairability were criteria assessed in previous trade studies, including the Composite Payload Fairing Structural Architecture Assessment and Selection documented by Krivanek and Yount.9 That trade study rated damage tolerance equivalent for honeycomb and foam sandwich structures but did indicate that foam sandwich structures were less susceptible to fabrication defects. The Lightweight Spacecraft Structures & Materials (LSSM) project report entitled, “Ares V Interstage Structural Concepts Down Select Process and Results,” also indicated that, from a damage tolerance perspective, foam and honeycomb sandwich were equivalent. This report also rated the two cores as being equivalent from a repairability standpoint.

40 Inspection capability was also acknowledged as a key component for composite structures. MSFC Non-Destructive Evaluation personnel provided the assessments in table 20 with consultation from other Agency personnel. From this aspect, aluminum honeycomb possessed an advantage over foam core.

Table 20. OHC data.

Al Honeycomb Much more field experience than other candidate designs and proven methods exist that can be used right off the shelf. The aluminum provides a good contrast (attenuation) to voids (air). Foam Core The form core does not provide a good contrast (attenuation) to voids (air).

Discussion was held about the extensibility of the selection—whether or not the construction would be able to be utilized for structures on other parts of the vehicle, or for on-orbit application. This led to discussion of NASA-STD-6016, specifically section 4.2.6.2 which gov- erns Sandwich Assemblies.10 Section 4.2.6.2.c. of NASA-STD-6016 states that perforated and moisture-absorbing cores shall not be used in sandwich assemblies. This indicated that a Material Usage Agreement would have to be obtained for whichever construction was selected. A previous Advanced Composites Technologies task had performed a moisture-absorption test. The composites for that particular study did not have any surface treatment on the skins. That study indicated that honeycomb sandwich would lead to less moisture update than the foam core tested.

Another area of extensibility considerations or its own unique criteria was response to vibro-acoustic loading. The previously referenced Composite Payload Fairing trade rated the foam core as possessing better performance as compared to honeycomb. However, the ratings were very close for the two core types.

Another consideration for the wall construction trade was the ability to incorporate penetration and secondary attachments. The Interstage Trade Phase 1 Down Select FOM results had rated honeycomb as preferable to foam core for this consideration. However, discussions for the CEUS project resulted in concurrence for the condition that the methods for local reinforcement were well understood for sandwich construction, regardless of core type.

The following schedule and cost impact differences were also considered for the trade:

• Each unique core segment will require a design model: – Element properties will be adjusted for different densities of cores. • If the buildups are to the inside, this could require machining of the core, which increases in difficulty with decreasing density of core. • Segment-to-segment honeycomb cell communication may require stints or slots of cores that require additional panels and coupons to demonstrate viability of method and any impact to properties. • Methods of joining foam could require development. • Ensuring proper conditioning of foam to obtain a successful core to facesheet bond could require development.

41 • Core splice of honeycomb could cause facesheet issues and requirement development panels • Core dimpling of facesheets may require limited development to determine optimal cure pressure.

The Interstage trade had indicated the total cost for a foam core structure would be less than a honeycomb structure. However, for this project, both sandwich construction methods were determined to meet the cost and schedule constraints to produce a skirt STA.

One of the driving requirements from the CEUS Systems Requirements document states the assembly, including planned penetrations and interfaces, shall be at least 20% less massive than an analogous metallic skirt.

In the absence of meaningful differentiation between the sandwich configurations, this requirement was the driving factor in the recommendation to select the core type. The only mass data obtained during the wall construction trade, as outlined in the analysis section (sec. 4), was for a condition limited to one penetration. One of the concerns cited for composite structure implementation is that during preliminary sizing, activities composites offer significant weight advantages; but once penetrations and attachments are included, the mass savings are not as substantial as originally anticipated. Due to the limited time to conduct the wall trade and design as well as build and test a STA from the selected configuration, there exists some possibility that the mass growth of one configuration over the other could have brought the numbers within a range where factors, including cost and schedule, may have been influential in selection of the appropriate core type. However, this could not be assessed for this project; thus, honeycomb sandwich was selected as the wall configuration for the remainder of the STA activities on the project. Local reinforcement would be traded as the design matured as to utilize high-density aluminum core or potted aluminum core, or high-density foam core or inserts.

Additionally, it should be noted the Ares V Interstage report that had stiffened composite structure offers the potential to minimize mass for certain launch vehicle applications. During the timeframe of the Wall Construction trade and post SRR, it was assumed that a design could revisit other configurations for the skirts that may lead to optimal mass savings, but that exceeded CEUS program constraints.

3.3 Design Requirements for Structural Test Article

The method adopted for structures design tasks by the Structural & Mechanical Design Branch at MSFC is to have a document referred to as a design data sheet (DDS) that lists criteria that must be met for the structure. Other design activities have similar products, where the criteria for determining the design are listed in a design and analysis notebook. The formal documentation for the criteria will be the Engineering or Element Requirements documents and/or NASA standards invoked in the signed stress report that accompanies the release of the drawings or delivered at the product critical design review. Due to the desire of this activity to use the loads applicable to the actual EUS, the DDS is marked as an International Traffic in Arms Regulation (ITAR) document. Figure 22 is the cover sheet included with the document.

42 Figure 22. DDS for the forward skirt STA.

The method of configuration management for this project was to utilize a Microsoft Share- Point site hosted by MSFC with limited access folders to provide configuration control. This document, however, is a preliminary document and was thus kept in the design folder. Configuration management for the DDS was to be by listing the date at the end of file name to provide a means of version control. The DDS for the target application at the completion of the program are included in appendix D––Design. All of the loads and information that would render the document to be marked with ITAR have been removed.

3.4 Development of the Skirt Structural Test Article Layout

The flange-to-flange heightor f the skirt STA was assumed to be 70 in. The OML of the article was derived based on multiple criteria. First, the vertical leg of the metallic end rings was designed so that the centerline closely aligned with existing test simulator dimensions. From this constraint, the OML of the composite section was developed based on a thickness of core that would satisfy a load set that could be considered representative of vehicle loads for the targeted components. The OML was determined based on the honeycomb sandwich design as selected in the wall construction trade. Honeycomb core properties considered conducive to manufacturing were utilized in determining the thickness of the core.

43 Based on the CEUS-PROG-02—the assembly shall include interfaces representative of those in the SRR SLS/EUS baseline design, the following list of penetration and attachments were developed for the skirt STA layout:

• Forward skirt panel penetrations: – Umbilical plate (electrical, LH2 vent/relief, purge). – LH2 vents (two propulsive, two relief) (opening w/ gasket/seal). – Systems tunnel/panel. – Vents (if not provided by payload above), (opening w/ gasket/seal). – Cameras/fairings (mounting/cable penetration).

• Forward skirt panel secondary mounting/loads: – Flight computers (shelf/mounts). – Antenna (mounting). – Purge vent ducts. – Press line mounts. – Harness mounts. – Sensors (mounting/opening). – Ground support equipment hardware mounts.

• External interfaces: – Forward ring flange. – Aft ring flange.

It was decided that only the major penetrations would be represented in the STA. Secondary hardware was to be left to a decision of whether it was to be conducted on a panel level type test or assessed analytically. The STA layout (figs. 23 and 24), was thus generated as a baseline for detailed design to be initiated and to facilitate discussions with assembly fixture design and test hardware planning. The layout was distributed to the team through SharePoint.

44 Figure 23. Forward skirt STA layout.

Figure 24. Forward skirt STA layout second angle.

45 3.5 Evaluation of Model Based Design Tools

One of the objectives for the CEUS was to examine efficiencies related to model based design. Model based design has many variations and can range from no production of 2D products, e.g., drawings, to an approach where models are controlled but drawings are still utilized for integration, communication to multiple disciplines, and as a basis for quality control or inspection. If the models are controlled, then they can be distributed to Analysis or Manufacturing, but a system must be in place to maintain configuration control if the models are modified from the original computer-aided design (CAD) format. A proposal for the definition of ‘model based’ for this project was presented to the MSFC Design Skill Leads as well as submitted to the CEUS project management. This definition, as included in the SRR package, was ‘the design model will be analyzed for producibility utilizing manufacturing simulations prior to design approval. The approved design model should also provide sufficient definition of ply geometry to ensure corresponding structural analyses can be iterated efficiently with a maturing design.’

3.5.1 Evaluation of Alternate Tools

The CAD software primarily utilized at NASA for hardware design is Creo from PTC. The companion product data management system Windchill has an instance for NASA specified as ICE Windchill. The configuration and data management processes for flight hardware development are based on this suite of tools and reflected in the corresponding quality instructions of the Agency. The standalone Creo tool, however, did not provide for manufacturing simulations of a composite article, nor did it provide an automated manner to address ply drops. For manufacturing simulations, the decision was made to use CGTech VERICUT for Composite Paths for engineering (VCPe), a companion software to the CGTech VERICUT VCP used to drive the automated fiber placement (AFP) robot possessed by both MSFC and LaRC. Fibersim, a CAD supplemental package supplied by Siemens, was utilized for ply drop definition and to facilitate manufacturing drawing generation.

In addition to the baseline suite of tools that required the VERICUT and Fibersim aug- mentations listed above, a fully integrated solution provided by Dassault Systèmes was evaluated. The functionality between the two toolsets was compared, and for this activity, the augmented PTC Creo capability was selected. This decision was based on the fact that flight hardware design processes at MSFC are currently being tailored to the PTC toolset and the familiarity of the NASA user community.

3.5.2 Approach to be Utilized to Enable Models to Serve as Basis From Design to Manufacturing

The following approach was established in order to meet the definition for model based approach for this project as defined in the SRR package. Models are generated using Creo CAD software (fig. 25). The Fibersim supplemental package is used within Creo to generate ply information, including definition of ply drop locations (fig. 26).

46 Figure 25. Creo model of tool surface.

Figure 26. Fibersim is used to generate plies.

Fibersim generates cross sections of plies that can be used by the designer to visualize ply drops (fig. 27). Generated cross sections are also used in manufacturing drawings. The cross section is a drawing detail that is required to properly communicate the ply configuration by defining the boundaries for every ply dropoff or buildup area. Depending on the complexity of the design, cross-section views could take an extensive amount of time to generate manually using Creo without the Fibersim supplemental package. With Fibersim, the views can be generated in a matter of minutes. Furthermore, Fibersim is designed to allow ply definition and ply cross sections to be easily updated as the design evolves (fig. 28).

47 Figure 27. Fibersim generated 3D cross section of laminate showing interleaved plies.

Figure 28. Fibersim has a ply sequence editor that allows the designer to easily edit the ply sequence.

48 When the design is finalized, the Creo model and the Fibersim ply data are used to generate manufacturing simulations to assess the manufacturability of the part. The Creo model and the Fibersim ply data are imported into VERICUT VCPe (fig. 29). Within VCS, the model and ply data are used to generate manufacturing simulations (fig. 30). That same information can be used in CGTech VERICUT VCP to generate course data for the AFP machine (fig. 31).

Figure 29. CGTech VERICUT VCPe with imported Creo model and Fibersim ply data.

49 Figure 30. CGTech VCS manufacturing simulation.

Figure 31. CGTech VERICUT VCPe-generated course data for AFP (reinforcement ply).

50 3.6 Composite Universal Stage Adapter

Upon completion of the CEUS skirt layout and prior to the detailed design progressing to the PDR, the project received direction to ensure that the products had extensibility for SLS hardware development. Due to budget and time limitations, panel test options for SLS hardware risk reduction were considered. However, it was decided to focus on the USA2, a structural component of the SLS forward of the EUS. The baseline USA2 design was sandwich composite with aluminum honeycomb core, and was divided into four cone-cylinder petals that were connected with metallic, vertical separation joints.

A trade study was conducted to determine the most applicable test within the budgetary constraints of the project. The scope of the trade extended from a STA that replicated the structure of a flight USA2 (with a slightly shortened barrel that would be constructed using the tool procured for the CEUS skirt) to testing a peta––a one-fourth section of the cone and cylinder. Also considered within the trade was a test of either just the conic or just the cylinder sections of the USA2. The configurations considered in the trade are shown in figure 32.

Figure 32. CUSA2 STA traded assembly options.

51 3.7 Structural Test Article

The trade for the USA2 had a comprehensive scope that led to the selection of a cylindrical test article utilizing the cylindrical tool already procured during the CEUS skirt section of the project. The CEUS had higher loads than the USA2, so the CEUS tool was designed for a thicker core than the USA2 required. This resulted in offsets of the neutral axis between the existing test simulators and the test article. The end rings had to be designed to account for this offset. The STA would be between 12 and 13 ft tall with an outer diameter, D, of 331.2 in, and it would include one 36-in penetration positioned at least 1D from the top of the panel. The STA would be constructed of eight sandwich composite panels to be joined with OOA, out-of-oven vertical joints. Excluding the metallic vertical separation joints from the STA reduced complexity. At the time of this decision, a trade was being conducted as to whether vertical separation joints were required for the flight USA2. The panels would have intermediate modulus/toughened epoxy facesheets bonded to an aluminum honeycomb core using epoxy. The metallic end rings would be a separable clevis design with eight segments joined together with splice plates. The STA would be manufactured using the AFP machine to lay up the facesheets of the eight panels. The panels would be placed in the autoclave to cure the facesheets. An assembly fixture would be used to support the formation of the OOA, out-of-oven vertical joints connecting the panels.

Dynetics was contracted to design the assembly fixture for the CUSA2 STA. Interface information was provided to Dynetics for the STA interface with the test simulators. Dynetics began to use this information to design an assembly fixture interface that would use the same bolt hole pattern as the test simulators. Figures 33 and 34 depict the drawing created to communicate this interface. (All ITAR information has been removed.)

52 Figure 33. Interface definition for the STA assembly fixture (sheet 1).

Figure 34. Interface definition for the STA assembly fixture (sheet 2).

53 To meet the requirement of testing the STA to a representative flight load case, the loads were reviewed for the DAC-0 Block 1B Cargo and Crew Configurations. An ascent load case was chosen for the STA to facilitate using the existing ET test simulators without structural modification. No burst or crush pressure loads would be tested because the complexity of pressurizing the test setup was a cost and schedule risk to the project. The test would be conducted at an ambient temperature. A DDS was generated for the STA that summarizes the material, construction, and load information. The geometry of the STA was defined in a layout drawing (shown in figs. 35–37). (All ITAR information has been removed.)

Figure 35. Layout drawing for the STA (sheet 1).

54 Figure 36. Layout drawing for the STA (sheet 2).

Figure 37. Layout drawing for the STA (sheet 3).

55 3.8 Pathfinder Design Effort

In addition to the STA, a Pathfinder design effort was proposed that would essentially conduct a PDR for CUSA2, with the assessment of the product limited to an analytical assessment. The Pathfinder would be designed to encompass all of the load cases for the DAC-0 Block 1B Cargo and Crew Configurations. Figure 38 is an illustration of the conceptual model design. The Pathfinder load cases would also incorporate the crush and burst pressure loads. All load cases including stiffness requirements are documented in the Pathfinder DDS.

Figure 38. Pathfinder conceptual model.

Trades planned for the Pathfinder study included a trade of the OML shape of the structure. In this trade, a two-part cone and cylinder construction would be compared with alternate geometries that incorporated a smooth, curved transition between the cone and the cylinder por- tions of the USA2 (fig. 39). Factors such as manufacturability, mass impact, and aerodynamic properties would be considered in the trade. Other trades that were planned within the Pathfinder effort were a core material and density study, knockdown factor evaluation based on imperfection sensitivity, a vertical separation joint trade, and a discontinuity factor sensitivity study.

56 (a) (b)

Figure 39. OML trade study considerations: (a) Two-part cone and cylinder and (b) alternate geometries.F39_1624

For the Pathfinder effort, transient temperature profiles would be developed for hot/cold prelaunch (DSNE) terrestrial environments (e.g., rollout, on-pad, tanking) as well as ascent aero- thermodynamic environments. It was determined that the initial case would assume no Thermal Protection System (TPS). Aerothermal heating data were not available for USA2 in the Block 1B DAC-0 database, so the Thermal Branch (EV34) planned to coordinate with the Aerosciences Branch (EV33) to determine reasonable aeroheating assumptions for the preliminary assessment.

57 4. JOINTS AND ANALYSIS OVERVIEW

4.1 Composite for Exploration Upper Stage Skirt Activities

The CEUS project began with a focus on the skirts that provide connections between the service module and the LH2 tank for the EUS at one end, and between the LH2 tank and the Interstage at the other end. In addition to the normal launch and ascent loads, the skirts experience significant thermal gradients because of their proximity to the cryogenic propellants.

4.1.1 Wall Trade Analysis

The first trade study carried out for the CEUS program was designed to determine an appropriate wall construction for a composite skirt STA designed to launch vehicle requirements. Included in the trade were the following construction types:

• Aluminum honeycomb core sandwich. • Foam core sandwich. • Corrugated panel. • Blade stiffened skin. • Hat stiffened skin.

These composite construction concepts were sized for the SLS loads and environments and compared against metallic orthogrid and isogrid concepts. Ground rules and assumptions, as well as trade study results, are included in the wall construction trade study.

4.1.1.1 Models. A finite element model (FEM) of the hydrogen tank aft skirt was developed utilizing eight panels joined with longitudinal joints (because of the physical limitations of the autoclave that was to be used). This skirt model was then attached to the load introduction structure at the fore and aft of the skirt to simulate the structure at these locations and provide appropriate stiffness (fig. 40). The attachments at these locations were made with representative metallic clevis joints. Additionally, a 12-in segment of metallic structure (a simple ring) was included to provide separation between the composite part and the extreme temperatures seen at the tank interface (fig. 41). Finally, a penetration to the primary structure (for human access) was also modeled to understand the impact of this design detail on the results.

58 Load Introduction Forward Ring (AI) Forward Buildup Structure (AI) Forward Clevis (AI)

Longitudinal Joint (AI)

Acreage

Acreage Buildup Aft Buildup Penetration Aft Ring (AI) Load Introduction Aft Clevis (AI) Structure (AI)

Figure 40. Aft skirt sizing model. F40_1624

Forward Ring (AI) 12 in Clevis FWD Composite Sandwich Acreage

Composite 70 in 54 in Buildup at Sandwich Opening Acreage

Al Insert at 30 in Opening Aft Ring (AI) 4.75 in 4 in 331 in

Figure 41. Joint interface details.

F41_1624

59 4.1.1.2 Sizing Results. The skirt sizing was performed with HyperSizer and NASTRAN as the analysis tools. The sizing results are shown in table 21. Several important results came out of the sizing results. The mass of the interface rings and joints were much more than the acreage masses of any of the concepts. Additionally, all of the sandwich constructions ended up at an assumed minimum gauge thickness (8 plies) as the principal failure mode was buckling. This allowed for some assumption of damage tolerance, as the design was nearly identical for pristine material properties or open-hole compression equivalent material properties. It was initially assumed that the stiffened designs might prove to be more efficient than the sandwich constructions. However, the large thermal gradients at the tank interface caused a bi-axial stress state that was more effectively reacted by the sandwich constructions. There were several design concepts that met the required 20% mass benefit compared with the metallic baseline.

Table 21. Wall construction trade study mass results.

Aft Skirt Acreage Buildup Adhesive Total Mass Savings Mass Mass Mass Mass Buckling (Metallic Baseline) Damage Tolerant Material Properties (lb) (lb) (lb) (lb) Eigenvalue (%) Honeycomb core (3.1 pcf core) 398 123 62 1,580 2.50 42 Honeycomb core (4.5 pcf core) 474 150 62 1,683 2.84 38 Honeycomb core w/penetration (4.5) 462 191 62 1,712 3.01 39 Foam core 637 195 62 1,891 2.85 31 Foam core w/penetration 632 280 62 1,971 3.00 30 Corrugated core 634 165 31 1,828 3.07 33 Blade stiffened 1,451 253 16 2,717 2.24 1 Hat stiffened (2-in hats) 970 205 16 2,187 2.16 20 Hat stiffened (4-in hats) 842 195 16 2,050 2.21 25 Pristine Material Properties Honeycomb core (4.5 pcf core) 556 147 62 1,700 – 38 Corrugated core 634 137 31 1,800 3.08 34 Blade stiffened 1,451 253 16 2,717 2.24 1 Hat stiffened 780 148 16 1,941 2.48 29 Metallic orthogrid 1,345 710 – 3,052 2.22 –12 Metalic isogrid 1,442 296 – 2,735 3.30 – Metallic isogrid–w/penetration 1,472 332 – 2,802 2.47 – Note: Metal frame weight = 997 lb.

4.1.1.3 Figures of Merit Evaluations. Other figures of merit used to determine the preferred acreage construction are documented in section 4.2.1.1.

60 4.1.2 Joint Trade Study

Following the wall construction trade study, a second trade study was undertaken to determine the appropriate joints for the STA. The STA required eight longitudinal joints because of the facility limitations. Circumferential joints were also evaluated at the forward and aft interfaces. The joint concepts evaluated included the following:

• Longitudinal joints: – Double-lap, composite bonded joint. – Double-lap, composite bolted joint. – Double-lap, metallic bolted joint. – No longitudinal joint (remove these altogether—allow separation of panels).

• Circumferential joints: – Metallic clevis-bolted joint. – Metallic clevis-bonded joint. – 3D woven F-preform joint (bonded). – 3D woven Pi-preform joint (bonded). – Builtup F-preform joint (bonded).

Masses for each of the joint concepts were evaluated with spreadsheet calculations as well as open-source code (A4EI) for the adhesive connections. Additionally, the concepts were evaluated in other key areas, in much the same way the wall construction was evaluated. These key areas were:

• Structural integrity. • Inspectability. • Minimum recurring cost. • Minimum nonrecurring cost. • Minimal development cost. • Design/analysis uncertainty. • Producability/complexity.

The joint trade study was not completed prior to the program request to consider alternate hardware.

4.1.3 Kissing Bond Evaluation

A kissing bond describes a region where substrates intended to be bonded in contact are held together by a very weak bond, but cannot sustain tensile or shear stresses required by the design. A kissing bond is the limiting case of a weak bond (i.e., a very weak bond that will likely fail under service loads). The likelihood of a kissing bond occurrence can be minimized using manufacturing process controls since kissing bonds result from manufacturing process errors. For composite sandwich structure, a kissing bond at the facesheet-core interface is a relevant failure mode that must be considered in structural certification. In this work, the effect of local disbonds

61 on the strength of sandwich structure is considered. This work complements a similar investigation into the effect of facesheet-core disbonds on buckling of sandwich panels.11

Structural substantiation of disbonds can be accomplished using a damage tolerance approach. For a damage tolerance approach, a critical flaw size and detectable threshold must be established. A margin of safety is established by ensuring that the critical flaw size is at least two times larger than the detectable size.

In this section, a preliminary damage tolerance evaluation of facesheet-core disbonds in composite sandwich structures is described. Several analyses were conducted using Abaqus Stan- dard version 6.14. Material property data available in the literature were used. The geometrical configuration, facesheet layup, and materials were assumed based on guidelines from the CEUS trade study activity, which were conducted in parallel. Though the particular model configuration does not match the outcome of the CEUS panel sizing, the configuration used here is representative and the model is parametric so accommodating different configurations in future analyses is simple.

No experimental validation was performed. Further work, including detailed experimental validation, is required to establish confidence in the analysis results described in the following subsections.

Section 4.1.3.1 describes investigation into the critical size of facesheet-core disbonds subjected to relevant loading conditions, and section 4.1.3.2 discusses two concepts for disbond detection.

4.1.3.1 Critical Size. For the purpose of structural substantiation, kissing bonds can be treated as disbonds since they effectively have no strength. Structural test and analysis of facesheet- core disbonds has been conducted in prior investigations. For example, Rinker12 and Glaessgen13 have used the virtual crack closure technique (VCCT) to analyze facesheet-core disbonds in the presence of internal pressure. In order to use a fracture mechanics approach such as the VCCT, the fracture toughness of the facesheet-core interface must be characterized. Ratcliffe presents a recommended procedure for characterization of facesheet-core fracture toughness and have tabulated the fracture toughness of the facesheet-core interface for several facesheet and core materials from data available in the literature.14 In this project, a 3D parametric model was developed by building on the previous work referenced above. The purpose of the model is to evaluate the effect of a facesheet-core disbond on the strength of sandwich structure subjected to edgewise compression load. The details of this analysis are described in this section.

4.1.3.1.1 Model Configuration. The model configuration was chosen to resemble the standard test for edgewise compressive strength of composite sandwich construction (e.g., ASTM C364), as shown in figure 42.

62 Facesheets

= 7.9 Core

Ny y

t f w = 5.9

tc = 1 Figure 42. Configuration used for analysis of the critical size of kissing bonds at the facesheet-core interface. F42_1624

The key difference between the configuration shown in figure 42 and the standard test is that a circular disbond is considered with diameter, D. The disbond is placed at one facesheet-core interface. The configuration has a width,w, of 5.9 in and a height, h, of 7.9 in. The overall dimensions were chosen so that the behavior of the disbond is not influenced by edge effects. The core and facesheets have thickness, tc=1 in and, tf = 0.057 in, respectively. A python script was used to automatically generate models with various disbond sizes, which will greatly assist future studies.

Since the primary load for the CEUS structures is axial compression, it is assumed that a compressive load acts on the model in the y-direction as shown in figure 42. Under this edgewise compressive load, the driving force for initiation and propagation of the disbond is local buckling of the debonded region. Therefore, a postbuckling analysis procedure was used with two steps. In the first step, a linear eigenvalue buckling analysis is conducted. The first buckling mode (local buckling of the disbond region) is used to seed an imperfection in the mesh for the subsequent analysis with an amplitude of 0.03 tf for the second step. In the second step, a geometrically nonlinear postbuckling analysis procedure is used where the edgewise compressive load is applied as a prescribed end shortening and cohesive elements or the VCCT are used to capture growth of the disbond and thus predict the strength of the specimen. This analysis procedure extends the work of Reeder et al. to the case of sandwich constructions. 15

63 4.1.3.1.2 Mesh and Boundary Conditions. A typical mesh is shown in figure 43. The core and facesheets were represented with one layer of continuum shell elements (SC8R) each. A refined mesh was used in the region near the disbond front and a coarser mesh was used in the far-field region. The typical far-field mesh size was 0.25 in. For models that used cohesive elements, the mesh size in the refined mesh region was 0.01 in to satisfy guidelines for cohesive element size.16 The cohesive elements (COH3D8) were placed in the region highlighted in green in figure 44. The cohesive elements were replaced with tie constraints during the first analysis step (eigenvalue buckling analysis). For models that used VCCT, a typical refined mesh size of 0.08 in was used, as established by a mesh convergence study. For both models with cohesive elements and models that used the VCCT, a contact condition was used in the second step to prevent the facesheet from penetrating the core within the disbond region. Also, in both models, damage growth was only allowed within the refined mesh region. Once damage propagation reached the transition region, the model results were truncated, since damage is artificially arrested at this point the results are no longer physically meaningful.

Disbond Region Refined Mesh; Damage Growth Transition Region

Disbond Front

Disbond

(a) (b)

Figure 43. Typical mesh used for the critical disbond size analysis: (a) Front and (b) side. F43_1624

64 y

uy = uy at y =

u = 0 on the Perimeter

( = 0, y = 0, = w, y = ) on the = 0 Surface

u = uy = u 0 at = y = = 0 uy = 0 at y = 0

Figure 44. Boundary conditions for the critical disbond size analysis. F44_1624

4.1.3.1.3 Material Properties. Material properties for the facesheets, core, and facesheet- core interface were assumed based on published data summarized in NASA Internal Memo- randum, Property Values for Preliminary Design of the Ares I Composite Interstage, J. Reeder, 2007).The facesheets are assumed to be IM7/8552 with a 0.0057-in ply thickness and a [45/–45/02/ 902/02/–45/45] layup. The core material was assumed to be Rohacell® 110 HERO foam. Core elastic properties for Rohacell HERO foam is E = 27 ksi and v = 0.3. Mode-independent damage propagation was assumed for the facesheet-core interface using a Mode I value for fracture toughness, which is a conservative assumption since Mode II and mixed-mode fracture toughnesses are typically higher than the Mode I fracture toughness. The fracture toughness was assumed as 5.71 lbf/in.14 For the cohesive elements, the strength was assumed as 1,827 psi,16 and the penalty stiffness was calculated following the recommendation by Turon et al.17

65 4.1.3.1.4 Results. A typical contour plot of uz from the eigenvalue buckling analysis for the first eigenmode is shown in figure 45 where the disbond size is 1.6 in.This deformation is used to seed an imperfection in the subsequent postbuckling analysis.

u 1

Disbond Front

–0.02

Figure 45. Typical first mode of eigenvalue buckling analysis. F45_1624

Typical load-displacement responses, damage propagation, and deformation are shown in figure 46 for a model with cohesive elements and in figure 47 for a model with VCCT. In both cases the initial disbond size is 1.6 in. In figure 46, three key points in the load-displacement response are designated as follows: damage initiation, I; onset of load-displacement nonlinearity, N; and peak load, P. The damage state and out-of-plane displacement are shown as contour plots for each key point in the load-displacement history. It is observed that damage has initiated and propagated to a critical length when the peak load is reached. Similarly, in figure 47, the peak load point is highlighted, and corresponding contour plots of the damage state and out-of-plane displacement are shown. The VCCT prediction does not indicate damage initiation prior to the peak load.

66 Initiation (I) Initiation (I)

Nonlinear (N) Nonlinear (N)

Damage Index 10 P N 8 I Peak (P) 1 0.047 Peak (P) 6

4 0.5 Ny (kip/in) 0.021

0 0 –0.006 0 0.01 0.02 0.03 0.04 0.05 (a) (b) End Shortening (in) (c)

Figure 46. Damage, load-displacement response, and out-of-plane F46_1624 deformation predicted using cohesive elements: (a) Damage propagation, (b) load displaced response, and (c) uz (in).

10 Peak (P) Peak (P) 0.05 8

6 0.02 (kip/in) y

N 4 –0.01

2 Intact Failed 0 0 0.01 0.02 0.03 0.04 0.05 (a) (b) End Shortening (in) (c)

Figure 47. Damage, load-displacement response, and out-of-plane deformation F47_1648 predicted using VCCT: (a) Damage propagation, (b) load displacement response, and (c) uz (in). 67 In contrast to the prediction using cohesive elements, the model with VCCT does not capture damage initiation and initial accumulation, but it does capture damage propagation at peak load. The differences between the predictions from the two fracture mechanics approaches are expected due to the fact that the VCCT approach is based on a linear elastic fracture mechanics formulation while the cohesive elements consider both strength and fracture through a nonlinear fracture mechanics formulation. While the model using the VCCT provides less insight into the damage process compared with the model using cohesive elements, the model using the VCCT requires only 20% of the computational time of the model with cohesive elements. The difference in computational expense between the two models is due to (1) the difference in mesh size and (2) the cohesive element model suffering from more extensive load-increment convergence difficulties than the VCCT model.

Both models require the analyst to specify solution parameters that can have a significant impact on the results. For the VCCT model, a contact stabilization parameter of 1×10–6 was used as recommended by Krueger.18 Trial and error showed that larger values of this parameter and automatic selection of this parameter by Abaqus yielded varying results that were in poor agreement with the predictions from the model with cohesive elements. For the model with cohesive elements, viscous regularization was used to limit the convergence difficulties. A viscous regularization coefficient of 1×10−6 was found to provide a good balance between improving convergence behavior and having a minimal effect on the load-displacement response. Increasing the viscous regularization coefficient resulted in a load-displacement behavior with a rounded peak load instead of a sharp load drop. Models with smaller values of the viscous regularization coefficient converged significantly more slowly and had little effect on the load-displacement response. Though both solution parameters (contact stabilization parameter and viscous regularization coefficient) are the same, this is likely a coincidence for the particular structure considered here. It is assumed that the solution parameters established here are valid for the range of model configurations considered and that vastly different structures will require different solution parameters.

The disbond size was varied to obtain the load-carrying capability as a function of disbond size for this sandwich structure. The results are shown in figure 48 as predicted by the linear eigenvalue buckling analysis, the postbuckling analysis using cohesive element, and postbuckling analysis using VCCT. The results show that the models using cohesive elements and VCCT are in good agreement for the range of disbond sizes considered. The eigenvalue buckling analysis under- predicts the load-carrying capability when the disbond is large. However, the eigenvalue buckling analysis is useful as a quick and conservative preliminary design tool for assessing the severity of this failure mode.

68 10

6 (kip/in) y

N 4

Cohesive 2 Eigenvalue Buckling VCCT 0 0 0.5 1 1.5 2 2.5 3 3.5 4 D (in)

Figure 48. Critical disbond size.

F48_1624

4.1.3.1.5 Conclusions and Recommendations for Future Work. A parametric FEM was developed to assess the critical size of facesheet-core disbonds in sandwich structures subjected to edgewise compressive load. Two fracture mechanics-based approaches were considered to model propagation of the disbond: the VCCT and cohesive elements. Material properties available in the literature were used for the analyses. Model results show good correlation between the two modeling approaches. As a result, the use of the VCCT is recommended for future evaluations due to its lower computational cost. The results suggest that, for the CEUS design load levels, facesheet-core disbonds must be very large (≥4 inches) before this defect poses a threat to the structural configuration considered.

Detailed experimental characterization of material property inputs and validation of model predictions has not been conducted and is a required next step. In addition, the parametric model developed here could be used to develop critical disbond size data for a wide variety of sandwich configurations. Such data would be valuable tools for assessing the severity of the facesheet- core disbond failure mode in preliminary design. Further investigation for particular structural configurations including the effect of curvature, combined loads, environment, disbond shape, and disbond proximity to structural features such as joints are needed to assess the severity of this damage mode fully.

4.1.3.2 Detection. Detection of kissing disbonds is challenging since conventional NDE methods (ultrasound and thermography) identify defects through the presence of voids and, by definition, kissing bonds do not have a void. Therefore, conventional NDE techniques are not capable of detecting kissing bonds and weak bonds. However, if the assumption is that the kissing bonds are localized and have no strength, they can be detected using a nonconventional approach, where the sandwich structure is inspected while subjected to a pressurization load so that the structural deformation reveals the presence of a kissing bond by its lower local bending stiffness. This approach may be suitable for identifying very weak bonds since the loading could lead to failure in the weak region. However, in this study, the approach is considered for disbonds

69 only. Two applications of this approach are reviewed in this section: (1) A commercially available product for local inspection called the elasticity laminate checker (ELC) and (2) a newly proposed wide-area inspection method.

4.1.3.2.1 Elasticity Laminate Checker. The ELC was developed by Airbus19 as a local inspection method to identify facesheet-core disbonds in sandwich structures. Figure 49 depicts a schematic of the device. The ELC operates by applying a vacuum to a local region on one side of a sandwich panel and measuring the relative displacement of the skin. The skin will deform significantly more in areas where disbonds are present since the bending stiffness of one facesheet is much less than the bending stiffness of the sandwich. The ELC will only detect facesheet- core disbonds on one side at a time. To have confidence that the structure has no facesheet-core disbonds, inspections must be conducted on all regions (including both sides) of the sandwich panel. The ELC is a commercially available product that has been validated and used in field applications by Airbus.

Displacement Measurement Vacuum Gauge Vacuum

Facesheets

Core

Figure 49. Schematic of the ELC.

F49_1624

4.1.3.2.2 Vacuum-Assisted Wide Area Inspection for Facesheet-Core Disbonds. While the ELC is a useful tool for local inspections, it is time-consuming to scan large areas. Using a similar approach, a vacuum-assisted, wide-area inspection concept is proposed. The concept is illustrated in figure 50. The panel to be inspected (shown with black solid lines) is installed on a vacuum fixture (shown with blue dashed lines). The air within the enclosed region is evacuated. Following the concept of the ELC, localized deformation in the out-of-plane direction will occur in regions where facesheet-core disbonds are present. The vacuum loading condition should be designed such that weak bonds fail and, thus, are readily detectable while nominal bonds are undamaged. With the vacuum applied, detection of disbonds can be accomplished through any technique that allows for full-field measurement of out-of-plane deformation including using a coordinate measuring machine (CMM), shearography, or digital image correlation. In contrast to the ELC, the vacuum- assisted, wide-area inspection concept allows for rapid inspection of large panels.

70 CMM

Vacuum Chamber

Sandwich Panel

Figure 50. Wide-area, vacuum-assisted disbond inspection method. F50_1624

4.1.3.2.3 Detectable Size. The model for eigenvalue buckling analysis described in section 4.1.3.1 was used to evaluate the detectable size for the particular combination of geometrical configuration, layup, and material properties that were assumed. The deflection,δ , as a function of disbond size for vacuum pressure of 14.5 psi, is shown in figure 51. These results suggest that the threshold of detectability of facesheet-core disbonds is on the order of 2 in and 0.75 in for CMM and shearography based systems, respectively.

0.015

0.010 δ (in)

0.005

CMM

Shearography 0 0 1 2 3 4 D (in)

Figure 51. Detectable size for vacuum-assisted disbond inspection.F51_1624

71 Considering the analysis described in section 4.1.3.1, which indicates that the critical disbond size is on the order of 4 in, the ratio of critical size to detectable size is 5.3 for a system using shearography, which compares favorably with the typical minimum acceptable value of 2. There- fore, these analysis results suggest that a damage tolerance approach to mitigating the risk of failure due to disbonds at the facesheet-core interface is feasible.

4.1.3.2.4 Conclusions and Recommendations for Future Work. The feasibility of detecting disbonds via local and wide-area, vacuum-assisted inspection methods was assessed. Considering a representative sandwich structure configuration, an FEA suggests that disbonds as small as 0.75 in are detectable for flat panels. Further work is required to explore the feasibility and limitations of the proposed approach for kissing bonds with varying strengths through experimental tests and additional analyses. Future tests and analyses should consider the effect of curvature carefully.

4.2 Composites for Exploration Upper Stage Universal Stage Adapter Activities

Following direction from the Space Technology Mission Directorate (STMD) and the SLS program, the focus of the CEUS program shifted from the skirts on the EUS to the USA. The USA provides an interface between the EUS and the MPCV. Additionally, it serves as a fairing for a secondary payload (fig. 52).

Co-manifested Payload & Secondary Payloads

Baseline USA Option 1 USA Option 2 USA (Separating) (Separating) (Nonseparating)

Figure 52. USA options. F52_1624

72 4.2.1 Pathfinder Analysis

As the CEUS program began to consider the alternate geometry of the USA, initial sizing was completed to determine a point of departure for the flight-like construction of the USA. This became known as the Pathfinder design.

4.2.1.1 Initial Concepts. The CUSA Pathfinder Initial Concept design consisted of a conical portion and a cylindrical portion with a metal ‘hip joint’ clevis-type fitting to connect the cone to the cylinder. The entire assembly was divided into four petals that used a composite honeycomb core sandwich construction. The four petals were connected to each other using separation joints with clevis-type fittings to connect to the composite sandwich panels. The assembly was attached to the forward and aft structure using circumferential separation joints with clevis-type fittings.

The FEM was constructed using Patran (fig. 53). The model consisted of shell elements to represent the composite sandwich panels, metal separation joints, and other adjacent structures. Beam elements were used to represent the clevis fittings. The aft end of the model was restrained in the vertical and radial directions. The flight loads were applied as component loads to the forward end of the model and to the top of the cylinder using RBE3 connectors, as these do not add artificial stiffness. The flight loads used are the USA2 Loads – Crew Configuration dated December 18, 2014.

Figure 53. Analysis model for USA initial design concept.

Linear static analyses were performed on the FEM using NASTRAN. The structure was optimized for weight using a combination of HyperSizer and NASTRAN. Linear buckling analyses were then performed on the optimized structures.

This model was used to provide an initial core thickness and layup of the cone and cylinder acreage panels. Once the Alternate Geometry CUSA model was constructed, the initial model was no longer used.

73 The initial USA design (conic section on top of the cylindrical section) came about because of a requirement to use the conic system separately (called USA1 at the time). However, this requirement was removed, and the system was evaluated in the baseline configuration (shown in fig. 54).

(a) (b)

Figure 54. USA: (a) Baseline geometry and (b) representative buckling results. F54_1624

4.2.1.2 Alternate Geometry. The CUSA Pathfinder Alternate Geometry design consisted of a conical portion, a cylindrical portion, and a blend portion that allowed for a smooth transition from the cone to the cylinder. The entire assembly was divided into four petals that used a composite honeycomb core sandwich construction. The four petals were connected to each other using separation joints with clevis-type fittings to connect to the composite sandwich panels. The assembly was attached to the forward and aft structure using circumferential separation joints with clevis-type fittings.

The FEM was constructed using Patran, shown in figure 54. The model consisted of shell elements to represent the composite sandwich panels, metal separation joints, and other adjacent structures. Beam elements were used to represent the clevis fittings. The aft end of the model was restrained in the vertical and radial directions. The flight loads were applied as component loads to the forward end of the model and to the top of the cylinder using RBE3 connectors as these do not add artificial stiffness. The flight loads used were the USA2 Loads – Crew Configuration dated 12/18/2014.

74 Linear static analyses were performed on the finite element model using NASTRAN. The structure was optimized for weight using a combination of HyperSizer and NASTRAN.. Linear buckling analyses were then performed on the optimized structures. Table 22 shows the resulting weights and buckling load factors (eigenvalues) of CUSA structures with varying facesheet layups, core thicknesses, and shell buckling knockdown factors (SBKFs).

Table 22. Buckling trade study results.

Core Thickness Weight Cone- (in) (lb) Cylinder Cone Blend FS Blend Lower Plies SBKF Limit* Cyl. Cone Blend Frame Panels Diff Mode Eigne Mode Eigen Mode Eigen 7 0.33 0.500 0.500 0.625 0.500 1113 2827 – 1 2.477 102 5.278 173 6.149 7 0.65 0.500 0.500 0.500 0.500 1113 2778 –49 1 2.477 60 4.635 >400 >8.3 6 0.33 0.250 0.500 0.625 0.500 1113 2530 –297 1 2.161 121 4.643 >400 >7.4 6 0.65 0.250 0.500 0.375 0.250 1113 2406 –421 1 2.155 34 3.082 5 2.378

* Self-imposed lower limit of the core thickness range (core thickness may not be optimum).

4.2.2 Structural Test Article

The CEUS project investigated the potential of building and testing a test article that would represent all or a portion of the USA. After several iterations, it became clear the program would be unable to fund anything larger than a cylindrical section of the full USA article. As a result, some modifications to the sizing needed to be made to ensure that the test article would be representative, given facility limitations in manufacturing and in testing. The subsequent design was called the STA.

4.2.2.1 Initial Sizing. The STA composite laminate layups were based on the sizing for the Pathfinder cylindrical portion. The core thickness was based on the wall thickness evaluation and the load-carrying ability of the existing LOX and LH2 simulators (interface test hardware from previous programs).

4.2.2.2 Wall Thickness Evaluation. The STA consists of a 12-ft-tall version of the cylindrical portion of the CUSA. The assembly consisted of eight composite honeycomb core sandwich panels joined longitudinally using composite laminate doublers on both the interior and the exte- rior. Coarse versions of the LOX and LH2 tank simulators were used to emulate realistic boundary conditions for the STA (fig.55). The STA connects to the forward LOX tank simulator and aft LH2 tank simulator using metal clevis-type fittings.

75 Figure 55. STA model with typical buckling eigenmode response.

The FEM was constructed using Patran. The model consisted of shell elements to represent the composite sandwich panels, composite laminate doublers, metal clevis fittings, and metal simulator structure. Beam elements were used to represent the simulator ring frames and vertical hat stiffeners. The doublers and clevis fitting were connected to the composite sandwich panels using glued contact. The aft end of the model was restrained in all directions. The flight loads were applied as component loads to the forward end of the model using an RBE3 connector, as it does not add artificial stiffness. The flight loads used are the Ascent Bin 8 – Max Q (Cylinder Aft) minus the shear load.

Linear buckling analyses were performed on the STA FEM with various core thicknesses, as the capability of the test stand hardware was being evaluated in parallel. Figure 56 shows the core depth versus the buckling load; this figure was used to estimate the core thickness required to ensure failure of the test article prior to the test facility hardware.

76 4.5 4.34 4

3.5

2.5 2.37

Buckling Load Factor 2

1.5 1.40

1 0 1/8 1/4 3/8 1/2 5/8 Core Depth (in) F56_1624 Figure 56. STA core depth study results.

4.2.2.3 Simulator Evaluations. The STA along with the test facility hardware (called simulators) and load ring models were analyzed to evaluate the buckling load of the STA and, most importantly, the strain distribution in the simulators. Based on an initial assessment of simulators by Dawn Phillips (fig. 57), the allowable line loads in the LO2 and LH2 simulators were limited to 3,300 lbf/in and 3,700 lbf/in, respectively. These line loads would result in an approximate factor of safety on yield of 1.1. Later assessment of the simulators based on modified failure criteria generated allowable line loads of 6,200 lbf/in and 4,900 lbf/in for LO2 and LH2 simulators, respectively. In order to estimate the load distribution in the simulators during actual testing of the STA in the full stack configuration, a transient dynamic analysis of a full stack model (consisting of load rings, simulators, and STA) was performed to evaluate buckling and postbuckling behavior of STA and also to obtain strain and displacement distribution at the interfaces of simulators and STA. The high-fidelity FEMs of the simulators and the loading rings were developed using shell elements in which frames and stringers were explicitly modeled. The full stack model (fig. 58) included loading jacks used in the simulation of load application points.

77 Figure 57. Simulator capability memorandum.

78 Loading Jacks

Loading Ring (Top)

LOX Simulator

STA

LH2 Simulator

Loading Ring (Bottom)

Figure 58. Full stack model—assembly of STA, simulators, loading rings, and load jacks. F58_1624

A sandwich core thickness of 0.5 in was used (and fixed) in the STA model to evaluate the simulators. Nodes in the flange region of the bottom load ring were fixed in all degrees of freedom and a uniform axial displacement was applied to the top nodes of the loading jacks. The load versus displacement curve and radial displacement of the STA are presented in figure 59.

(×1E6) 3.5 U, U1 (CID1) 3 +3.183e+00 +2.336e+00 2.5 +1.490e+00 +6.435e–00 2 –2.030e–00 –1.049e+00 1.5 –1.896e+00 Load (lb) –2.742e+00 1 –3.589e+00 0.5 –4.435e+00 X –5.281e+00 0 Y –6.128e+00 –6.974e+00 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 End Shortening (in)

Figure 59. Load versus displacement curve and radial deformation contours in stack.F59_1624

79 The radial deformation contours in the LH2 simulator (to assess bending behavior of the simulator flange) are shown in figure 60. This initial assessment showed a test line load that was very close to the allowable limits for the simulators so a slightly different deformation configuration of load jacks (fig. 61) was considered to decrease the line load in the simulators. This deformation configuration resulted in generating axial and bending moment loads in the ratio as defined in Bin8. The load versus axial displacement of the STA and its radial deformation for displacement configuration defined above is shown in figure 62. This load configuration resulted in one side of the STA deforming more than the other. The radial deformation contours, before and after complete buckling, are shown in figure 63.

U, U1 (CID1) U, U1 (CID1) +4.066e–02 +1.199e–02 +3.495e–02 +8.014e–02 +2.924e–02 +4.041e–02 +2.353e–02 +6.794e–02 +1.783e–02 –3.905e–02 +1.212e–02 –7.878e–02 +6.413e–03 –1.185e–01 +7.061e–04 –1.582e–01 –5.001e–03 –1.980e–01 –1.071e–02 –2.377e–01 –1.642e–02 –2.774e–01 –2.212e–02 –3.172e–01 –2.783e–02 –3.569e–01

(a) (b)

Figure 60 Radial deformations in LH2 simulator: (a) Just before buckling,F60_1624 and (b) at the end of the analysis.

–0.4635 –0.059

–0.75 0.225

–0.4635 –0.059

Figure 61. Axial displacement of load jacks (in inches). 80

F61_1624 (×1.E6) 1.2 Max Load=1.177 E+6 lb U, U1 (CID1) 1 1.730e+00 1.205e+00 6.795e–01 0.8 1.543e–01 –3.709e–01 0.6 –8.960e–01 –1.421e+00 0.4 –1.946e+00 Axial Force (lbf) –2.472e+00 0.2 –2.997e+00 X –3.522e+00 Y 0 –4.047e+00 –4.572e+00 0 0.05 0.10 0.15 0.20 0.25 Average Displacement (in)

Figure 62. Load versus displacement curve and radial deformation contours in stack. F62_1624

U, U1 (CID1) U, U1 (CID1) +5.675e–01 +1.730e+00 +4.953e–01 +1.205e+00 +4.230e–01 +6.795e–01 +3.508e–01 +1.543e–01 +2.786e–01 –3.709e–01 +2.063e–01 –8.960e–01 +1.341e–01 –1.421e+00 +6.189e–02 –1.946e+00 –1.033e–02 –2.472e+00 –8.256e–02 –2.997e+00 –1.548e–01 –3.522e+00 –2.270e–01 –4.047e+00 –2.992e–01 –4.572e+00 (a) (b) F63_1624 Figure 63. Radial deformation contours in the STA: (a) Just before buckling and (b) just after buckling.

In figure 64, the maximum in-plane, principal-strain distribution in the LH2 simulator just before and after buckling is presented. Since strain values are not verified for convergence, they should be used for qualitative purposes only.

81 E, Max. In-Plane E, Max. In-Plane Principal SNEG, Principal SNEG, (fraction = –1.0) (fraction = –1.0) (Avg: 75%) (Avg: 75%) +1.833e–03 +1.860e–03 +1.666e–03 +1.675e–03 +1.500e–03 +1.491e–03 +1.333e–03 +1.306e–03 +1.166e–03 +1.122e–03 +9.993e–04 +9.375e–04 +8.325e–04 +7.530e–04 +6.657e–04 +5.685e–04 +4.990e–04 +3.840e–04 +3.322e–04 +1.995e–04 +1.654e–04 +1.500e–05 –1.324e–06 –1.695e–04 –1.681e–04 –3.540e–04

(a) (b)

F64_1624 Figure 64. Strain contours in LH2 simulator: (a) Just before buckling and (b) just after buckling.

There was still some residual concern that the test facility would be unable to test the STA to failure without damaging the simulators. A nonconventional approach to buckle the STA was explored to assess the performance of the full stack model without damaging the simulators. In this configuration, all of the load jacks were uniformly deformed by 0.75 in, and a transverse load of 1 kip was applied normal to the surface and in middle of the longitudinal joint as shown in figure 65. The transverse load resulted in an inward radial deflection of 0.19 in at the load application point. The intent was to ‘trip’ buckling failure prior to the nominal buckling load.

Loading Jacks

–0.75 –0.75

–0.75 0.75

1 Kip (0.19-in Radial Deflection)

–0.75 0.75

–0.75 –0.75

Figure 65. Displacement and load configurations (in inches). F65_1624

82 This load configuration, like the previous one, resulted in one side of the STA experiencing more compression than the other side and resulted in nonuniform buckling of the STA as shown in figure 66. In figure 67, radial displacement of the STA before and after buckling is presented.

(×1.E6) 2.5 U, U1 (CID1) +1.904e+00 2 +1.331e+00 +++7.572e+01 +1.836e+01 1.5 –3.899e+01 –9.635e+01

Load (lbs) 1 –1.537e+00 –2.111e+00 –2.684e+00 0.5 –3.258e+00 X –3.831e+00 0 Y –4.405e+00 0 0.1 0.2 0.3 0.4 0.5 –4.978e+00 End Shortening (in)

Figure 66. Load versus displacement curve and radial deformation contours in stack.F66_1624

U, U1 (CID1) U, U1 (CID1) +3.459e–01 +1.904e+00 +2.670e–01 +1.331e+00 +1.881e–01 +7.572e–01 +1.092e–01 +1.836e–01 +3.035e–02 –3.899e–01 –4.852e–02 –9.635e–01 –1.274e–01 –1.537e+00 2.063e–01 –2.111e+00 –2.851e–01 –2.684e+00 –3.640e–01 –3.258e+02 –4.429e–01 –3.831e+00 –5.218e–01 –4.405e+00 –6.006e–01 –4.978e+00

(a) (b)

Figure 67. Radial deformation contours in the STA: (a) Just before buckling F67_1624 and (b) just after buckling.

83 Table 23 shows the buckling load, line load, and the in-plane principal strain in the LH2 simulator for all three load configurations analyzed in this study. The in-plane principal strain should be used in assessing changes to the strain distribution in the simulator when the load configuration varied among the three load cases.

Table 23. Summary of buckling load line load and principal strain.

Max Principal Strain Buckling Load At Buckling Line Load Type of Loading (lb) (μ) (lb/in) Uniform axial displacement 3.46 E+6 1,549 3,327.35 Bending moment 1.18 E+6 1,833 1,134.76 Uniform axial displacement 2.55 E+6 1,232 2,454.24 with 1 kip transverse load

4.2.2.4 Buckling Load Sensitivity to Geometric Imperfections Study. The STA along with the simulators and load ring models were analyzed to evaluate the sensitivity of the buckling load of the STA to manufacturing geometric imperfections. Axial displacements were applied to the load jacks as to simulate the Ascent Bin 8 (Max Q) loads (fig. 68). Transient dynamic nonlinear analyses were performed on a STA perfect geometry model and a STA model with imperfections that were scaled and superimposed on the nodal coordinates of the STA. The imperfections were taken from a composite 8-ft barrel mandrel from Griffin Aerospace. The magnitude of the imperfections was 1X the measured imperfections from the Griffin Aerospace 8-ft mandrel as shown in figure 68. In reality, the actual imperfections of a STA if manufactured could vary from the assumed imperfections in this study.

0.120 0.105 0.089 0.073 T8–001 0.057 0.041 0.025 0.009 –0.007 –0.023 20–001 –0.039 –0.055 –0.070 –0.086 Y –0.102 –0.118 X

Figure 68. Nodal imperfections applied to the STA.F68_1624

84 The load-deflectioncurve from the transient dynamic nonlinear analyses for the perfect STA and STA with imperfections is shown in figure 69. The perfect STA model a load ratio of 3.91 for Fz and 3.93 for My at buckling. The STA with imperfections has a load ratio of 3.62 for Fz and 3.64 for My at buckling which is a reduction in the buckling load by 7.4%. Figures 70 and 71 show the radial deformation for the perfect STA and STA with imperfections, respectively at the buckling load and the last load increment in the transient dynamic analysis.

1,400,000 Buckling Loads • STA Perfect Model 1,200,000 CUSA2 STA Perfect Model – F = –1,175,952 lb, M = –197,797,855 in-lb CUSA2 Model With Imperfections y – F Load Ratio = 3.91, M Load Ratio = 3.93 1,000,000 y

800,000 • STA With Imperfections – F = –1,090,978 lbs, My = –186,476,071 in-lb 600,000 – F Load Ratio = 3.62, My Load Ratio = 3.64

400,000

Average Applied Displacement (in) Average 200,000

0 0 0.05 0.1 0.15 0.2 0.25 0.3 Average Applied Displacement (in)

Figure 69. Load deflection curve of transient dynamic nonlinear analysis.F69_1624

U, U1 (CID1) U, U1 (CID1) +8.672e–01 +1.808e+00 +7.238e–01 +1.283e+00 +5.804e–01 +7.572e–01 +4.370e–01 +2.316e–01 +2.936e–01 –2.940e–01 +1.502e–01 –8.196e–01 +6.763e–03 –1.345e+00 –1.366e–01 –1.871e+00 –2.800e–01 –2.396e+00 –4.234e–01 –2.922e+00 –5.668e–01 –3.447e+00 –7.102e–01 –3.973e+00 –8.536e–01 –4.499e+00

XY XY

(a) (b)

F70_1624 Figure 70. Radial deformation of perfect STA model: (a) At buckling load and (b) at last load increment in the transient dynamic analysis.

85 U, U1 (CID1) U, U1 (CID1) +8.052e–01 +1.708e+00 +6.720e–01 +1.207e+00 +5.388e–01 +7.068e–01 +4.055e–01 +2.062e–01 +2.723e–01 –2.944e–01 +1.391e–01 –7.950e–01 +5.858e–03 –1.296e+00 –1.274e–01 –1.796e+00 –2.606e–01 –2.297e+00 –3.938e–01 –2.797e+00 –5.270e–01 –3.298e+00 –6.603e–01 –3.799e+00 –7.935e–01 –4.299e+00

XY XY

(a) (b)

Figure 71. Radial deformation of STA model with imperfections: (a) At buckling load and (b) at last load increment in the transient dynamic analysis.F71_1624

A comparison of the buckling load ratio and its corresponding compressive line load are listed in table 24, as well as an eigenvalue buckling analysis and the nonlinear transient dynamics analysis for the perfect STA model and STA model with imperfections.

Table 24. Summary of buckling load sensitivity to geometric imperfections study.

Compressive Line Load Model Buckling Load Ration (lb/in) 0.5-in sandwich core eigenvalue buckling analysis 4.03 3,522 0.5-in sandwich core transient dynamic analysis—perfect 3.91/3.93 3,426 0.5-in sandwich core transient dynamic analysis—imperfection 3.62 3,178

4.2.3 Trade Studies

4.2.3.1 Core Density (4.5 pcf versus 3.1 pcf). Initially, a honeycomb core with a density of 4.5 lb pcf was selected for the STA in order to simplify manufacturing. The thin foil used in some lighter core constructions can prove difficult to handle. However, there are opportunities to achieve mass reductions for the virtual flight model (VFM) using lower density core than the core planned for the STA. The Core Density trade study was an attempt to understand the impact of this type decision on the performance of the VFM.

Manufacturing large panel segments calls for handling 96-in×48-in aluminum honeycomb core segments. This can lead to core damage. The large spacecraft structures and materials manufacturing effort used 3.1-pcf-density core with 0.125-in cell size and 0.0007-in foil thickness. The big core segments with the thin foil damaged easily, and required lots of inspection and core

86 edge straightening. To mitigate damage, a 4.5-pcf core with 1/8-in cell size, 0.001 in foil thickness was selected for the STA. To show the potential for mass improvements a 3.1-pcf core with 3/16 in cell size, 0.001-in foil thickness was selected for a trade study. This larger cell size could lead to print through and facesheet dimpling with the co-bonded manufacturing approach. This dimpling between larger cells has the potential to reduce ultimate buckling load (due to local imperfections) or to reduce facesheet strength. Table 25 shows the test matrix adopted to evaluate the buckling and strength concerns by using the larger cell size.

Table 25. Lightweight core validation test plan.

No. of Test Coupon Environment Coupons Edgewise Compression 3.1 pcf per drawing 1281253-001 RTD 5 (ASTM C364) Edgewise Compression 4.5 pcf per drawing 1281253-003 RTD 5 (ASTM C364) Buckling 3.1 pcf per drawing 1281251 RTD 5 Buckling 4.5 pcf per drawing 1281253 RTD 5

4.2.3.1.1 Panel Manufacturing. Two sets of panels (4.5- and 3.1-pcf core) were fabricated using the MSFC and LaRC AFP machines. The panels were of sandwich construction with 4.5- and 3.1-pcf, 1-in-thick perforated aluminum core (5052 and 5056) with ( 3/16 and 1/8 in) hexagonal cell sizes and 8-ply quasi-isotropic graphite/epoxy facesheets (IM7/8552-1) with (45°/0°/–45°/90°)s stacking sequence. The intent was to show equivalence between the two different tow widths on the AFP systems. At the end of the CEUS program, the MSFC panels were prepared in different samples to complete the testing, but budget and schedule constraints made it impossible to do the same for the panels fabricated at LaRC.

4.2.3.1.2 Panel Nondestructive Evaluation. The panels built at MSFC were verified by ultrasonic inspection. Inspection results showed a few indications of bad areas between facesheet and core; these areas were cut around and not part of the panel sections delivered for testing. The complete inspection report is given in appendix E––Manufacturing.

4.2.3.1.3 Pretest Predictions.

4.2.3.2 Test Articles and FEMs. The as-machined coupons from the parent panels were 40.2×6 in and 8.2×6 in for buckling and EWC specimens, respectively. The core material was removed from the ends (1.2-in deep) and EA 9394 was used to fill the 1.2×6×1-in empty space. The intent was to reduce the risk of crushing the core at the ends. After a full room temperature cure, the specimen ends were potted with EA 9394 epoxy grout, using 1-in-thick aluminum plates cut from 0.25-in-thick-wall extruded tubes. Prior to this, the specimen-end was centered in the slot and squared. The slots at the panel’s corners at the edge remained unfilled to form an open cavity in order to act as stress relief features. Finally, the specimen ends were machined flat and parallel.

87 Femap and NX NASTRAN were used for pre- and post-processing and to assess buckling (SOL 105) and strength (SOL 101) performance. Two-dimensional plate (PCOMP) elements were used to model the facesheets, while solid elements were implemented for modeling the core (four elements through the thickness) and potted ends. The in-plane mesh size was in the order of 0.2×2 in. Rigid elements were used at the top and bottom to apply boundary conditions and a compressive axial load. The models and end condition details are shown in figure 70.

Core properties came from the Hexcel data sheet, while facesheet properties are from table 21 (Pristine allowables) (J. Reeder, NASA Internal Memo, Property Values for Preliminary Design of the Ares I Composite Interstage, 2007).

The coupons (and FEMs) were identical for both the buckling and edge-wise compression test articles, except for the EWC coupons that were 32 in shorter. Figure 72 shows both buckling and EWC FEMs with end-fitting details.

Buckling Coupon

6 in

Edge-wise 40 in Compression Coupon

6 in 8 in X Y X Y

8 in

Stress Relief Al HC Core

4 in Core Fill (1.2 in) 1 in Core Fill (EA 9394) X Y 1.2 in 0.25 in Y Potting Compound (EA 9394) Stress Relief

Figure 72. Buckling and EWC and FEMs with end-fitting details. F72_1624 88 4.2.3.3 Finite Element Analysis (FEA) Results. Table 26 summarizes the predicted failure loads and end-shortening loads for both EWC and buckling specimens.

Table 26. Predicted loads and end-shortening: Core trade study.

Predicted load Predicted End- Core Density at Failure Shortening Test (pcf) (kip) (in) EWC 3.1 58.1 0.07 4.5 58.1 0.07 Buckling 3.1 36.7 0.25 4.5 38.4 0.26

Figure 73(a) and 73(b) also illustrate the buckling performance (SOL 105 results) for the 4.5 and 3.1 pcf core, respectively.

1 1 0.938 0.938 0.875 0.875 0.813 0.813 0.75 0.75 0.688 0.688 0.625 0.625 0.563 0.563 0.5 0.5 0.438 0.438 0.375 0.375 0.313 0.313 0.25 0.25 0.188 0.188 X 0.125 X 0.125 Y 0.0625 Y 0.0625 0 0 Output Set: Eigenvalue 1 3.843487 Output Set: Eigenvalue 1 3.672731 Deformed (1): Total Translation Deformed (1): Total Translation (a) Nodal Contour: Total Translation (b) Nodal Contour: Total Translation

Figure 73. Buckling performance: (a) 4.5 pcf and (b) 3.1 pcf core models.F73_1624

89 The facesheet, core, and core-fill strengths were also assessed at these buckling critical loads to ensure that the buckling will be the primary failure mechanism for these coupons. For instance, for the 4.5 pcf core, the facesheet failure index (FI), based on maximum strain failure criterion, at the buckling critical load of 38.4 kips is about 0.66. Realizing that FI ≥ 1 indicates failure, this FI (fig. 72) provides a healthy margin against facesheet strength failure at buckling. Core and core-fill stresses are relatively low and show no risk of failure prior to buckling.

For the EWC coupons, on the other hand, the primary failure mode is facesheet failure that occurs at 85.1 kips. Failure index contour of these specimens at the ultimate load indicates an FI = 1 (fig. 74). It should be noted that this FI is calculated based on pristine allowables of J. Reeder’s Memo, which carry some statistical basis. The actual strength value for the lamina (and facesheet) could be at a higher value.

0.662

0.635

0.608

0.582

0.555

0.528

0.501

0.474

0.448

0.421

0.394

0.367

0.34

0.314 X 0.287 Y 0.26

Output Set: NX NASTRAN Case 1 0.233 Deformed (0.262): Total Translation Elemental Contour: Envelope (Max Abs 1000090. 1000290. 1000490.10)

Figure 74. Facesheet failure indices at buckling. F74_1424

4.2.3.3 Test Results. Test results were not available in time for publication of this Technical Memorandum.

90 5. MANUFACTURING

Manufacturing for the CEUS project was intended to be a NASA internal effort. Section 5 will focus on the MSFC effort to mature, develop, and procure tooling for processes specifically tailored for highly optimized thin sandwich parts. Due to the aggressive accelerated schedule, the tooling procurement was initiated well before a preliminary design cycle was complete. While from a sched- uling viewpoint this appeared attractive, it presented several challenges and, ultimately, resulted in increased cost due to design changes early in the tool development.

A cure cycle effort was initiated to advance a cure cycle based on historical processes for 8-ply-thin laminates over a 3.1 pcf core. It was observed early in the project that the traditional cure cycle was resulting in areas of increased porosity. Historically, this has been observed with thin laminates and low-density core during co-cure processes. A multistep cure cycle approach was adopted to closely control the flow of the film adhesive and incrementally increase the pressure to result in a decreased amount of observed porosity.

The Manufacturing section will also cover the full size panel fabrication trials that were performed using the tooling package and cure cycles. Limited mechanical tests were performed before project cancellation that indicated the processes under development would have been suitable for the full-scale fabrication effort. The team was also in the process of developing a mature concept of operations for full-scale assembly. This effort was not fully matured and was left at the conceptual design phase. Further work is needed to define the assembly requirements and tooling necessary.

5.1 Tooling Development

The CEUS barrel concept specified a segmented panel arrangement. To facilitate this design required using a 1/8 OML-tooled, fiber-placement breakdown mandrel capable of withstanding 350 °F cures.

During manufacture of the tooling packages, numerous project reviews and technical coordination meetings were held to ensure functionality, schedule adherence, and cost validity. Typically, the following meetings were held for each tool:

(1) PDR held within 1 month of contract award to review and establish tooling concepts. (2) Final design review of preferred design specifics and provide authority to complete tool design. (3) Tool design buyoffs review completed design and provide authority to begin tool fabrication. (4) Physical tool buyoffs review fabrication and inspection data and provide authority to ship.

91 5.2 Composites for Exploration Upper Stage Tooling Requirements

The CEUS tooling activities are as follows:

(1) Facilitate the accurate creation of the CEUS within the stated tolerance range when cured at the specified 350 °F cure temperature. The concept for the CEUS Tooling Design is shown in figure 75.

Shackles for 4 pt. Pick (with spreader beam) • Weight: 3,000 lb • Max Width: 162 in Edge Bars 171 in • Overall-length: 20 in • Diameter: 331.79 in

65 in

Fork Tubes (both sides)

Hard Points/Adjustment Points FPM Interface Tow Bar Interface (both ends) Casters (removable)

Figure 75. CEUS tooling design concept. F75_1624

(2) Meet an as-fabricated profile tolerance of ±0.01 in. In addition to meeting the specified tolerances, the tool surface shall be smooth and fair with a surface finish of root-mean-square roughness of 32 μin or better on all surfaces within the edge of part (EOP).

(3) Include EOP scrible lines in layup tool.

(4) Include part number and measured weight in identification.

(5) Ensure tool is capable of tolerating 350 °F during cure without any mold degradation. This allows for temperature variations during the specified 350 °F cure cycle.

(6) Design tool with capability of withstanding, at a minimum, fifty 350 °F cure cycles, each cure cycle consisting of a dwell at 350 °F for 4 hr.

(7) Design tool with a minimum additional 12-in minimum run-out on all sides.

(8) Ensure tool is able to fit through an 18-ft-wide door and be capable of being lifted by a 5-ton crane with a single hook.

92 (9) Have lifting provisions that include four host rings on the backup structure for over- head crane operations and two pair of forklift tubes (4 in × 8 in × 60 in minimum rectangular steel tubes), 180° apart.

(10) Adhere to the following minimum factors of safety achieved throughout the tool’s life: • 5 FS on ultimate strength. • 3 FS of yield strength (metal parts only).

(11) Ensure tool can withstand the following transportation loads; loads are assumed to occur simultaneously in each of the three directions: • ±2 on fore/aft g’s. • ±2 on lateral g’s. • +3, –1 on vertical g’s.

(12) Provide work platform that will allow access to the tool surface while in the vertical orientation of the AFP cell rotators. Platform design shall meet OSHA (rails and toe boards and stability) requirements and be rated for 1,000 lb (load rating in pounds must be marked on platform).

(13) Construct tool surface of a carbon laminate. A polyimide topcoat to increase surface hardness and chemical resistance is required.

(14) Supply edge bars capable of fabricating two panels on the tool: • The edge bars shall have a cross section of 3 in wide and approximately 1.25 to 1.5 in thick. (The thickness will be finalized at the design and analysis review based on the finalized government design.) • The edge bars shall be chamfered to accommodate vacuum bagging. • The inside edge shall be located 2 in beyond the EOP. • Edge bars hall be divided into segments for ease of handling. • Tooling pins shall be located and sized to support a deflection of no more than 0.020 in at 80 psi autoclave pressure at the 350 °F cure cycle.

(15) Design tool to have vacuum integrity with a maximum leak rate of 1/4 in Hg in 60 min at 27 in Hg.

(16) Support tool substructure with at least four removable casters between 8- and 12-in-diameter phenolic wheels, swivel locks, and brakes.

(17) Supply three caul sheets, designed and fabricated to impart a smooth tool surface on the bag side of the part during the curing process: • The caul sheet shall be fabricated from carbon laminate and shall be sized to approximately 0.042 in. • The caul sheet shall be fabricated net to edge of part. • The caul sheet shall not require indexing scribe lines.

93 (18) Provide a tow hole at each end of tool, with an approximate 3.5-in diameter with steel- reinforcement plate to allow for transportation with a forklift or tug, as shown in figure 76.

Figure 76. CEUS final tool.

After arriving at MSFC, the tooling surface tolerances were verified using structured light scanning. The critical surface dimensions were within the specified tolerances, as shown in figure 77.

A112 D: –0.2232 0.5000 A108 0.4250 D: 0.0840 0.3500 A151 A100 0.2750 D: –0.0244 D: 0.1978 0.2000 0.1250 0.0500 –0.0500 A127 –0.1250 D: 0.0084 A093 –0.2000 D: –0.2591 –0.2750 A145 –0.3500 D: 0.0687 –0.4250 A130 –0.5000 D: –0.0042

Figure 77. CEUS metrology (dimensions are in mm). F77_1624

94 The CEUS tool was interfaced with the MSFC AFP cell (fig. 78) and located in the machine coordinate system. The trial for the CEUS tool AFP are depicted in figure 79. Programs were generated based on initial sizing. One of the major objectives of the CEUS project was to demonstrate accurate AFP over aluminum honeycomb core similar to what will be experienced at potential vendors.

Figure 78. CEUS tool installed in MSFC AFP cell.

Figure 79. CEUS tool AFP trials.

95 5.3 Assembly Concept

This section contains the project top-level goals centered around the development of equipment, tools, and processes necessary to support the manufacturing of the CEUS large composites structures. Specifically, the following is a list of initial efforts that will focus on developing techniques needed to manufacture the CEUS:

(1) Develop designs for tooling that support ease of manufacturing of the composites structures.

(2) Develop designs for tooling that is configurable to accommodate multiple composite panel configurations.

(3) Minimize operational complexity and cost. Design the tooling in such a way that frequent transfers between tools is not necessary.

(4) Integrate tooling with the current flow of operations that NASA uses in manufacturing the composite panels. This work will be an extension of the processes that are already in place.

(5) Begin scope of the tooling with receipt of a cured, trimmed, and machined panel. End scope of the tooling with a fully assembled interstage that has been transferred to a transporter.

The assembly tool is used during assembly of eight panels into one completed CEUS. This tool includes the following elements and capabilities, which are further refined in the CEUS Assembly Requirements document:

• Base and drive mechanisms: – Includes the base supports and the drive mechanism for rotating the CEUS. – The base supports are such that the lower edge of the aft ring are 3 to 6 ft above the floor surface, to allow personnel to have access to the aft ring during assembly operations. – The base has provisions for a smaller or larger barrel diameter.

• Drill station: – Supports drilling holes that are normal to the CEUS surface. – Performs match-drilling of aluminum flanges to composite panels. – Performs match-drilling of composite splice plates to composite panels. – Accommodates drilling of through-holes near the edges of each panel section. – Drills tower for indexing of holes in the vertical direction along the full height of the CEUS. – Tracks system for indexing the drill tower around the perimeter of the CEUS; limited to 1/8 or 1/4 of the CEUS circumference due to ability to rotate the entire assembly. – While in use, the towers should not restrict the ability of personnel to perform other assembly operations. – Drill station located such that bonding and drilling operations can occur simultaneously at two different joints (e.g., located 40 degrees from each other, or a multiple thereof).

96 • Bonding station: – Accommodate application of heat and pressure for curing joints and doublers, both sides of panel simultaneously. – Support pressure application up to 20 psi along entire joint height and across ~20 in of CEUS circumference at a time; pressure applied evenly along entire surface. – Support controllable heat application up to 250 °F (TBD) along entire joint height and across ~20 in of CEUS circumference at a time; heat applied evenly along entire surface. – Bonding station should be located such that bonding and drilling operations can occur simultaneously at two different joints (e.g., located 40 degrees from each other, or a multiple thereof).

• Tower sections: – Include provisions for precisely locating splice plates. – Are moveable to accommodate rotation of the CEUS assembly without interfering with other fixed assembly jig pieces. – Are moveable to accommodate removal of completed CEUS (if needed). – While in use, the towers should not restrict the ability of personnel to perform other assembly operations.

• Upper (forward) flange stiffening ring: – The upper stiffening ring attaches to the top of the vertical supports. – The upper stiffening ring includes a ring that accepts the forward flange pieces and has adjustments for locating the forward flange pieces.

• Lower stiffening ring: – Lower (aft) flange stiffening ring. – The lower stiffening ring attaches to the base and drive mechanism. – The lower stiffening ring receives the aft flange and has adjustments for locating the aft flange pieces. – The lower stiffening ring includes a rail system to allow rotation of the aft flange ring, which is itself connected to the vertical supports that are connected to the upper stiffening ring.

• Vertical supports: – The vertical supports interface between the lower stiffening ring and the upper stiffening ring. – The vertical supports include adjustments for the distance between the lower and upper stiffen ing rings and for the orientation of the lower and upper stiffening rings relative to one another (e.g., twist through the vertical axis). –Features.

• Access provisions: – The lower stiffening ring provides personnel access to bottom edge of the aft flange by providing 3 to 6 ft of clearance between the bottom edge of the aft flange and the facility floor. – Provide personnel access to top edge of the CEUS for installation of a ring. – Provide personnel access to inside of CEUS for section joining, etc.

97 • Sizing and adjustments: – 27.5-ft-diameter CEUS (nominal); with provisions for a minimum barrel diameter of 216 in and a maximum diameter of 330 in. – Approximately 13-ft-tall barrel section, including flanges, with provisions for accommodating a minimum barrel height of 60 in and a maximum height of 180 in. – Adjustments to bring the barrel sections into alignment for bonding and joining. – Interface constraints. – Interfacing with the facility beyond bolting equipment to the floor is not required. – Desire to not break the concrete floor and no concrete trenching for installation of fixture or equipment. – Ability to accommodate installation of a vertical separation joint is not necessary.

The assembly fixture concept and finished CEUS are pictured in figure 80.

Figure 80. Assembly fixture concept and finished CEUS.

5.4 Automated Fiber Placement Trials

Multiple large-scale composite sandwich panels were fabricated at MSFC using AFP in an effort to determine a variety of processing parameters associated with AFP and composite processing in general. The AFP parameters in question are compaction force, heater output, and feed rates. Other composite processing steps were developed, such as vacuum bagging sequences, core splice installation techniques, ply debulking frequency, and tool preparation details. The following list outlines the process developed at MSFC for fabricating large honeycomb core sandwich panels using AFP:

(1) Tool preparation is key to allowing parts to release from the tool easily and controls, to some extent, the surface finish of the parts. It was found that, for the composite tools, tooling sealer Chemlease MPP117 only needs to be applied after the tool undergoes a reworking which would involve thorough wiping with solvents and could include light sanding. Two coats of tooling

98 sealer were applied, allowing 30 minutes for each coat to dry. After the tooling sealer was applied/ dried, two coats of mold release, Frekote 700-NC, were applied, again allowing for 30 minutes between coats. This combination of initial tool sealer and mold release is sufficient for at least four autoclave cure cycles. After four cure cycles, two coats of NC700 should be applied. It was found that excessive use of tool sealer/mold release leaves a dull, milky finish on the composite parts. Also, only alcohol should be used to wipe the tool surface, acetone was found to be too aggressive of a solvent and required the reworking of tools.

(2) Once the tool is cleaned and prepared the first ply of composite 1/2( -in-wide IM7/8552-1 tow) was laid down using the AFP machine. It was found that the composite tape had sufficient tack to stick to the tool at ambient room temperatures without using a tackifier. For the first ply and solid laminates in general, a compaction force of 150 lb was determined to be sufficient. The heater was set to 150% for the first few plies; as the laminate becomes thicker, the heater should be reduced to 100%. As the laminate is being built, the tool/laminate starts to retain heat, and that increases the tack of the composite, which raises the tendency for tows to stick to the AFP compaction roller. A feed rate of 12,700 mm/min was used for the first ply to ensure the tows stick to the tool surface. For subsequent plies, the feed rate was increased to as much as 17,780 mm/min.

(3) After the first ply was laid, a debulk vacuum bag was installed to ensure the ply was adhered to the tool. The debulk should last for a minimum of 30 minutes. The debulk bagging scheme should follow the diagram in figure 81.

Vacuum Bag Breather Cloth Vacuum Sealant 1st Composite Ply Solid Release Tape Film

Tool

Figure 81. Debulk vacuum bag schematic. F81_1624

(4) After the first ply is debulked, the vacuum bag is removed, and the remaining plies can be fiber placed again using a feed rate of up to 17,780 mm/min, 150 lb of compaction force, and a heater setting of 100%–150 %.

(5) Once the first laminate is laid (fig. 82), the structural film adhesive (FM300-2) is installed onto the composite surface (fig. 83). The film adhesive is cut to the appropriate sized patterns using an automated Geber cutting table. The joints between adjacent film adhesive pieces should be overlapping a minimum of 0.25 in with a maximum overlap of 1 in. The overlapping of the adhesive ensures complete adhesive coverage of the composite part surface, while only sacrificing a fraction of 1 lb in ‘excessive’ adhesive.

99 Figure 82. AFP of first skin.

Figure 83. Film adhesive installed on first composite skin.

(6) Another debulk cycle is conducted as before to ensure the adhesive is firming adhered to the composite skin.

(7) The next step is the installation of the honeycomb core. The aluminum honeycomb core is cut/machined to size to ensure a tight fit at all core joints. There should be no gaps more than 0.1 in between adjacent honeycomb core pieces. It was found that just routine handling very easily

100 damaged the light density, 3.1-pcf aluminum honeycomb core. Special attention had to be taken to ensure cells were not deformed during handling/installation. It was also observed that the light density core exhibited areas of irregular cell shapes as seen in figure 84.

Figure 84. Honeycomb core cell irregular shapes.

(8) Next, the individual pieces of core were adhered together using a core splice adhesive. Two types of core splice adhesive were tried––a foaming core splice adhesive MA562, and a paste adhesive, EA9390. The foaming core splice adhesive was supplied in sheets 1/2 in thick, and 1-in- tall strips were cut to match the height of the core. The strips of foaming core splice are kept in a freezer until right before installation to keep the tack of the material as low as possible during installation. The installation of this type of core splice was simple, quick, and clean. When installing the foaming core splice, it is beneficial to work from one edge of the tool to the other, installing a piece of core/adhesive in the same manner bricks and mortar are laid. Figure 85 shows the foaming core splice installed on the edge of the honeycomb core.

Figure 85. Foaming core splice installation.

101 The EA 9309 paste adhesive was also used to join pieces of honeycomb. This paste adhesive is a two-part epoxy that was in gallon cans. The adhesive had to be mixed, pushed into cartridges, and injected into core joints using a pneumatic air gun (fig. 86). The mixing of the epoxy required the use of a vacuum mixer to ensure there were no air pockets in the adhesive during injection. The vacuum mixer is a large vacuum chamber that must be cleaned between each mix. The cleaning of the mixer was found to be quite laborious, and if more adhesive is required for a particular part than the vacuum mixer can accommodate in one run, excessive time will be lost to cleaning the equipment between adhesive mixes. It is recommended that, if a paste adhesive is desired, SIMCO kits are used instead. SIMCO kits will remove the need for a vacuum mixer, saving time. The joints were covered using 2-in-wide rubber strips to press the excessive adhesive that pushed up during the injection process. A vacuum bag was installed, and the paste adhesive was allowed to set up/cure for 12 hours.

Figure 86. Paste core splice adhesive injection.

(9) At this point, the second layer of film adhesive is installed onto the surface of the honeycomb core, in the same manner as the previous film adhesive layer, as shown in figure 87.

102 Figure 87. Second film adhesive layer with overlap.

(10) The film adhesive is then debulked down to the honeycomb for a minimum of 30 minutes. Before the debulking vacuum bag is installed, a core dam must be installed to prevent the crushing of the core on the edges of the panel.

(11) After removing the debulk vacuum bag, the second composite skin is fiber placed. The second skin is laid in a similar fashion to the first skin. The first ply has a lower feed rate, but in subsequent plies, the feed rate is increased. The heater output level starts at 150 % and is lowered to 100% as plies are laid. The main difference is the compaction is lowered to 50–100 lb. It was shown that no damage was induced to the honeycomb core cells from these compaction force levels as shown in figure 88.

Figure 88. View of honeycomb core after fiber placement of top skin, showing no damage.

103 (12) After the second skin is laid (fig. 89), the part asw bagged for autoclave curing following the bagging sequence in figure 90.

Figure 89. AFP of second skin onto surface of honeycomb core.

Vacuum Bag Porous TFP Breather Cloth

2nd Composite Ply Caul Sheet Vacuum Core Dam Sealant Honeycomb Core Tape 1st Composite Ply

Tool

Figure 90. Autoclave cure vacuum bag. F90_1624

104 (13) The part is then loaded into the autoclave and cured using the cure cycle in figure 91.

Figure 91. Part loaded into autoclave.

5.5 Curing

5.5.1 Scope

Optimal curing schedules are key to efficiently achieve the desired properties of the cured materials. Curing cycles determine the degree of cure of epoxy prepreg and have an important effect on the mechanical properties of the final products. Although companies manufacturing the commercial epoxy prepreg materials usually suggest curing cycles for custom applications, their curing cycles may not be the optimal ones for special applications. In order to optimize the curing cycles for epoxy prepreg used for the CEUS composite panels, it is necessary to understand the cure kinetics and characteristics of epoxy prepreg in more detail. To this end, the rheological properties of Hexcel 8552-1 prepreg and Cytec FM-300-2 film adhesive are characterized by means of dynamic mechanical analysis. This section will focus on the baseline cure cycle developed for the material systems used. All of the material had a nominal out time of 20 days. As discussed in section 2.2.4, the vilification and gelation temperature were affected with increased out-time.

5.5.2. Introduction

The flow behavior of a reacting system is closely related to the cure process. In the early cure stage, the epoxy resin is in a liquid state. Cure reaction takes place in a continuous liquid phase. In its liquid stage, the viscosity of a curing resin is influenced by two major phenomena. The first

105 is the increase in molecular size as cure advances, decreasing mobility, and hence increasing the viscosity. The second is the effect of the temperature on the mobility of these molecules. Resin viscosity thus depends on temperature and chemical conversion.

5.5.3. Gel Point

With the advancement of the cure process, a crosslinking reaction occurs at a critical extent of reaction. This is the onset of formation of networking and is called the gel point. At the gel point, epoxy resin changes from a liquid to a rubber state. Thus, after the gel point, the matrix can no longer flow and voids can no longer be suppressed. Although the appearance of the gelation greatly limits the fluidity of epoxy resins, it has little effect on the cure rate; so, the gelation cannot be detected by the analysis of cure rate, as is the case in the DSC. The gel time may be determined by a rheological analysis of the cure process. During a curing process under continuous sinusoi- dal stresses or strains, its viscoelastic characteristics change, which is reflected in the variations of rheological properties such as the storage modulus, E′; loss modulus, E″; viscosity, η; and loss tangent, tan δ. The E′ is the elastic character of the epoxy prepreg and reflects the energy that can be recoverable. The E″ represents the viscous part of the epoxy prepreg and reflects loss energy by dissipation. Viscosity measures the fluidity of the epoxy resin system or, more precisely, the frequency-dependent modulus. Higher viscosity means the lower fluidity of the epoxy resin systems. The loss tangent, tan δ, equals the ratio of the loss modulus to storage modulus. It is used to evaluate the viscoelasticity of epoxy prepreg. The gel point indication for DMA is defined by a sudden rapid increase in storage modulus or the maximum in tan δ.

5.5.4 Gel Point Cure Optimization

The optimal gel point should allow for the best flow characteristics with the least amount of voiding. Flow is dependent on the part’s geometry, size, and ply layup. For this reason, the optimal gel point may not be the same for every part but would use the same criteria for optimization.

5.5.5 Resin Flow

A controlled flow of the resin is desired during the initial stages of the cure. However, if pressure is increased too early during the cure, an excessive amount of resins bleeds out of the laminate layers leading to a void formation from a resin-poor region. When evaluating a cure cycle rheologically, a large drop in the storage modulus after autoclave pressure application is cause for concern. In this case, application of the full cure pressure is desired during an intermediate isothermal hold before the ramp to full cure temp at 355 °F. Thus, the isothermal hold period must be long enough and at a high enough temperature to allow for a sufficient viscosity increase. If a large drop in storage modulus is observed—half an order of magnitude or greater—during the subsequent temperature ramp to the curing temperature, the viscosity drop might induce poor resin distribution.

106 5.5.6 Volatile Removal

The removal of evolved vapor or air from the laminate is achieved by vacuum application during the cure, bagging schemes of the part, and the application of pressure. As temperature increases, solvent and water vapors are able to come out of the solution. If gelation occurs soon after a temperature rise without the external pressure required to suppress their formation, voids are induced into the matrix. These voids are homogeneous and spherical in nature (fig. 92).

Figure 92. Voids.

107 A higher viscosity reduces the efficiency of removing evolved vapors as flow is reduced. To improve vapor removal efficiency, a low temperature bake-out isothermal hold period is introduced at a lower resin viscosity.

5.5.6.1 Internal Stress. The DMA is recording the resistance of fiber movement in the matrix from a given strain. After gelation, the matrix is a single molecule and the orders of magnitude increase in storage modulus from gelation is a result of this consolidation into a singular solid rather than a solution of fibers in a viscous fluid. This transition to a rubber-like state induces stress that must be relieved by the matrix.

If gelation occurs during the heating ramp, the gel network must relieve the additional thermal stress induced by the thermal gradient within the composite. This stress relief can result in void formation and, in extreme cases, delamination between plies. This effect is scalar and is amplified by poor thermal transport through the part, which may restrict higher heating rates.

During an autoclave cure process, the maximum pressure must be applied before the gel point in order to suppress the formation of voids in the matrix but also have sufficiently high enough viscosity to resist bleeding of the laminate.

In order to achieve the optimal curing conditions, the full cure pressure is applied during an isothermal holding period at an elevated temperature for a given amount of time. The higher temperature removes more volatiles and increases the viscosity of the resin before pressure application.

The time to gelation for Hexcel 8552-1 prepreg is determined using a ramp rate of 1° F/min from room temperature. Run 1 simulates a 285 °F hold for 30 minutes cure cycle used in previous work. Run 2 is a 285 °F hold until the gelation and represents the longest time until gela- tin is achieved using the 285 °F hold period. Run 3 is a ramp to 355 °F with no intermediate hold period and represents the quickest time to achieve gelation using a ramp rate of 1 °F/min. Any hold period incorporated into the cure cycle will extend the amount of time required to achieve gelation but reduce the timeframe after the holding periods to bring the part to cure temperature before gelation.

As seen in the figure 93, the time to gel increased as anticipated in a roughly exponential fashion. However, gelation occurs during the ramp to the 355 °F cure temperature. Optimally, gelation of the laminate should occur after the heating ramp to reduce stress concentrations from fiber movement and rearrangement. The Run 3 DMA data indicate that even in the absence of a hold, gelation will occur during the ramp.

108 0.4

0.35 302.720753, 0.339895

0.3 292.840142, 0.28204

0.25 247.474699, 0.23867

0.02 Tan Delta Tan 0.15

0.1

Run 2.5 hour 0.05 Run 1 Run 3 Straight Ramp 0 0 100 200 300 400 500 600

Time (Min)

Figure 93. Tan delta peak at various 285 oF hold times. F93_1624

The DMA data in figures 94 and 95 suggest that a heating rate between 2 °F/min and 5 °F/min will achieve gelation during the 355 °F hold. The selection of heating rate must also factor in the thermal lag within the part. A higher heating rate increases thermal stress induced within the part. Thus, the objective is to have the slowest heating rate that can achieve gelation during the isothermal hold at cure temperature. A 2 °F/min heating rate is the lowest heating rate, which could achieve that condition.

109 200 0.3

180 0.25 160 172.714343, 0.226728 140 0.2 120 Tan 100 0.15 Δ

80 Temperature (C) Temperature 0.1 60

40 0.05

20 Sample Temperature Tan Delta 0 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400

Time (Min)

Figure 94. Ramp rate of 2 ºF/min with 30-min, 285 oF hold time.F94_1624

110 200 0.45

180 0.4

160 0.35

140 0.3 120

0.25 Tan 107.895084, 0.237586

100 Δ 0.2 80 Temperature (F) Temperature 0.15 60

40 0.1 Sample Temperature Tan Delta 20 0.05

0 0 0 50 100 150 200 250 300 0 50 100 150 200 250 300

Time (min)

Figure 95. Ramp rate of 5 oF/min to 30-min, 285 oF/min to cure temperature.F95_1624

5.5.7 Film Adhesive Considerations

5.5.7.1 Isothermal Hold at 250 °F. The CEUS composite panels are sandwich structures with composite laminate faces (Hexcel 8552-1) co-currently bonded to aluminum honeycomb core using an epoxy thermoplastic reinforce prepreg film adhesive (FM-300). The initial test panels showed evidence of resin bleeding through the film adhesive layer (fig. 96).

111 Figure 96. Micrograph of 8552-1 DDS/TGDDM resin (darker section) deforming the Cytec FM-300 film adhesive (lighter section) at core interface.

In order to control resin flow at the core interface, an elevated temperature hold period was incorporated into the cure cycle where the film adhesive can act as a barrier to resin flow into the core sections. Consequently, a lower temperate cure film adhesive, FM-300-2, was selected. FM 300-2 film adhesive is a 250 °F (121 °C) cure version of the widely used Cytec FM 300 film adhesive. The hold time and temperature were determined by the gel points indicated by the DMA data collected on the FM 300-2 adhesive.

As seen in figure 97, the gel point occurs shortly into the 250 °F hold. If the film adhesive does not achieve gelation relatively quickly into the hold period, the optimal conditions for laminate gelation and pressure application become more difficult to achieve. The gelled film adhesive can now act as a barrier layer and prevent excessive bleeding of resin into the core segment.

112 The second noticeable transition in figures 97 and 98 is the Tg transition occurring during the ramp to 355 °F cure temperature. Thus, the film adhesive is a gelled glass during the 250 °F hold and reverts back to the gelled rubber upon heating. This transition in the film adhesive does not significantly affect cure schedule considerations for the purpose of the present discussion, but is worth noting.

Loss and Storage Modulus Values of FM300–3 3 F Ramp Rate with 1 hr 250 F Hold 1.00E+12

185

165 1.00E+11 145

125 Tan 1.00E+10 105 Δ

85 Modulus Values (Pa *sec) Modulus Values 1.00E+09 65 Storage Modulus Loss Modulus 45 Series 1 1.00E+08 25 0 500 1,000 1,500 2,000 2,500 0 500 1,000 1,500 2,000 2,500

Time (min)

F97_1624 Figure 97. Loss and storage modulus values of FM300-3.

113 Tan Delta Peaks of 552 and FM300–2 3 F Ramp Rate with 1 hr 250 F 200 0.3 153.790577, 179.975 Gel Point 180 70.652155, 0.25045 Tg 0.25 160 134.266334, 0.239594

140 57.420062, 120.75 0.2 Gel Point 120 158.219083, 0.173855 118.450925, 121 Tan Δ 100 0.15

80 Temperature (C) Temperature 0.1 60 Sample Temperature Delta Tan 8552 40 Ten Delta FM 300–2 0.05

0 0 0 50 100 120 200 250 300 350 0 50 100 120 200 250 300 350 Time (Min)

F98_1624 Figure 98. Combining the results of the FM300-2 film adhesive and 8552-1.

5.5.7.2 Microscopy Evaluation

Microscopic visual inspection of distributed test panel cross sections reveals no evidence of delamination or elevated porosity. All cross sections were determined to have porosity lower than 1%. Some local porosity concentrations did exist but never in excess of 1%. A slight difference in porosity morphology was observed between the developmental cure schedules. Cure cycles using the low temperature of 185 °F hold developed pores that were typically smaller, more distributed with respect to ply thickness and spherical as shown in figure 99. Figure 100 shows that larger amorphous pores were present in cure cycles not incorporating the lower temperature volatile removal hold. Figure 101 shows a void likely caused by air entrapment.

114

Figure 99. Clean cross section using developed cure cycle.

Figure 100. Typical profile of porosity using developed cure cycle.

Figure 101. Example of void most likely caused from air entrainment.

115 This slight difference in the pore morphology suggests that the vapor phase did not coalesce and concentrate during the cure cycle, and air entrainment between plys was more efficiently removed with 185 °F debulking hold. This behavior seems preferable to the larger amorphous voids, which indicates a larger amount of stress within the composite laminate layer, as having an irregular shape indicates increased fiber rearrangement to relieve the vapor phase partial pressure contribution. Elongated regular voids occurring between ply interfaces typically indicate air entrainment.

There is no known study or data that suggests a decrease in the mechanical properties based solely on the void morphology but, larger defects typical require less energy to initiate failure modes than smaller distributed ones. Thus, the incorporation of a lower temperature volatile removal hold seems advantageous if the gel point still occurs during the isothermal cure temperature hold.

116 6. TEST

The USA2 STA had a two-part test plan. The first test of the STA would be to test the ultimate load in a manner to satisfy the project requirement of designing, building, and testing to conditions representative of flight conditions. The second test would be a test to failure. Both the ultimate load test and the test to failure would provide data that could be used to correlate with analytical models for validation or calibration. The SBKF team would conduct the test to failure. The SBKF study is an ongoing study aimed at developing SBKF test data and potentially lower- ing the conservatism from traditional design methods. The majority of the SBKF tests have been of metallic structures, so the CUSA2 STA allowed a unique opportunity to test a large composite structure to failure. For both tests, it was important to collect as much data as possible of the structure’s static and dynamic displacement and strain response to loading. Options considered for data acquisition were fiber optics and digital image correlation. Strain gauges alone would not collect the fidelity of data required to validate models. There was limited experience within the project with the fiber optics technology, and in recent implementation, there had been issues collecting the data, so the preferred method of full field displacement data acquisition was digital image correlation. Digital image correlation works by covering the test article with a high contrast speckle pattern and using specialized low-speed cameras to track the movement of the speckles. The tracked movement is then translated into displacement and strain information. An example of this is shown in figure 102.

4,000 Test Analysis W (mm) 3,000 Dynamic 30 2,000 Buckling

Load (KN) 1,000 20

0 10 0 500 1,000 1,500 Time (s) 0 –10 –20 –30 –40 –50

(a) (b)

Figure 102. Example of digital image correlation data: (a) Superimposed on a barrelF102_1624 that has been painted with a high contrast speckle pattern compared to (b) an analytical model of predicted displacement.

117 The two facilities considered for the STA test were Building 4619 and Test Stand 4699. Building 4619 was the favored option because it was indoors and allowed for the use of the digital image correlation cameras. Unfortunately, this facility had limited availability because SLS hardware had the priority. The second option, Test Stand 4699, was an outdoor test stand that made the use of the digital image correlation data acquisition system difficult. Figure 103 is an illustration of test configuration in Test Stand 4699. The outdoor use of the digital image correlation cameras was an issue because of environmental concerns such as wind, temperature, and lighting. Locating the cameras inside the STA was considered an option for mitigating the environmental concerns; however, this posed the risk of damaging the expensive cameras during the test.

Existing Hardware From ISPE Qual Test USA STA

Shear Beam Upper Spider

698.10 in Cylinder 13 ft 64 ft–9 3/16 in

650.82 in 46.24 ft

ALTA Simulator

Lower Spider

Figure 103. Test configuration in test stand 4699. F103_1624

118 The test simulators identified for the STA test were preexisting ET simulators. The simulators had a non-symmetrical bolthole pattern from their use in the ET tests. The “Interface Defini- tion (Sheet 1)” (figure 33) and the “Layout Drawing for the STA (Sheet 1)” (figure 35) was created to document this pattern. There was a concern that due to the factors of safety, the STA would not fail within the loads that the simulators could withstand. The ET simulators were analyzed to determine the maximum load they could safely tolerate. Contingency plans were considered for inducing failure in the test article if the maximum capacity of the simulators was predicted to be achieved prior to STA failure. The first option considered was introducing damage to the STA. This option was ruled out because, even with induced damage, there was a possibility that the STA would not fail within the limits of the simulators. Another option considered was applying a crush pressure to the test by creating a negative pressure within the STA. However, as mentioned previously, the complexity of pressurizing the test set up was a cost and schedule risk to the project, so it was decided that no burst or crush pressure loads would be tested. The contingency plan ultimately selected for failing the test article was to induce failure by applying a point load to the STA. While the STA was loaded in the axial direction, a lateral point load would be applied to the wall of the STA. This point load would be increased until the structure failed.

In addition to the STA, panel testing was also considered. A study was conducted to examine the feasibility of testing a 1/4 section of the cylinder and cone – referred to as a petal (figure 32 CUSA2 STA traded assembly options). In the feasibility study the petal had a metallic vertical separation joint. There existed a concern for the petal configuration that the stiffness of the joint would prevent the composite panel sections from receiving the proper loading. Additionally, the boundary conditions for the conic section were predicted to be critical for a stability test and assessed to be unfeasible to adequately replicate. It was determined that within the scope of this project cylindrical panels would be tested, but conic panels and petal sections were not an option. A plan was made to conduct cylindrical panel testing to obtain damage tolerance data in conjunction with the large STA structural test.

119 7. SUMMARY

This Technical Memorandum documented the work accomplished by the NASA CEUS team from when the project received authority to proceed until the project was discontinued due to the FY 2016 NASA Appropriations Bill impacts to the STMD budget. As documented, progress was made in maturing composite technologies, as shown in figure 104.

The project completed a successful SRR, selected and procured material, developed equivalency between panels manufactured at different centers, completed a wall construction trade study and selected preferred core configuration, performed initial analysis, matured NASA AFP capability, procured large composite tool, developed fabrication and processing parameters, and developed cure process optimization. The CEUS closeout documentation and data generated during the project will be stored by the Technology Demonstration Missions (TDM) program and retained on the CEUS Project SharePoint site.

Figure 104. Progress made in composite technologies.

120 APPENDIX A––LESSONS LEARNED

121 Comments, Risk, Closure ? I think that a NASA-wide spec for industry-standard products is imperative. Similar to Boeing’s BMS, it would give us a singular product that we vet once with the vendor. I think the dial-a-spec method we have now really confuses the vendor and somehow increases the production quality risk. We chose to use their internal spec alone, and it still took an extraordinary period of time and they had to make the For similar reasons as the spec, I also think some sort of a NASA-wide BPA (or similar) mechanism is needed for prepregs. Once we pick a resin/fiber combination-we have effectively sole-sourced it. Then we do the same dance; JOFOC, pricing, T&C again and again. Except fr more delays, none of this extra activity really affects the outcome. At our request, LaRC Procurement people are starting to take a look at this. We hope to My concern with our 145 gsm order was not just that Hexcel SLC made a bad batch, but that it left their plant in that state. Hexcel has to get their quality issues under control. I think marching up the food chain at Hexcel is the way to go for this. We should have one of our suits call one of their suits. This is supposedly aerospace grade material. All I heard from Hexcel SLC I think our order quantities may be a factor. It also could be that once we call them, they know we aren’t going to anyone else – so why hurry? I think standard NASA spec, pre-arranged procurement mechanism (BPA) and a little pressure from us will all help. [Justin] From a fabrication stand point we leveraged several years of development from CoEX, Ares and CCTD. We did not significantly learn any thing new past what was already developed. I believe the biggest challenges were going to be in the joining and assembly which we did not get to. Status Need date POC Sandi Miller / Brian Stewart Sandi Miller / Brian Stewart Sandi Miller / Brian Stewart Sandi Miller / Brian Stewart Sandi Miller / Brian Stewart Sandi Miller Justin R. Jackson Troy Mann Larry Pelham Implementation Plan / Action items Lessons Learned: Add a minimum of 4 months to the time you would expect for a similar procurement, and carefully review the quote. Make program management aware of the issues in dealing with the large prepreg suppliers; considering we buy relatively small quantities. Lesson Learned: The additional requirements are not acceptable to the vendor and slow the procurement process. If NASA would like to require specifications outside the internal spec, time should be planned to in essence develop a NASA spec. Lesson Learned: Try to work through this upfront with NASA procurement. I thought I had, but should have emphasized that the vendors will not change their policy. Lessons Learned: Make sure someone is on-site for at least the set-up of the prepreg run. Fortunately we were on-site and able to catch this issue early. Lesson Learned: Become familiar with the chain of command. I moved up that chain with both companies when response time became ridiculous. Relevance schedule Description Description: It took on average 3 to 4+ months to receive a quo te from Hexcel after the initial request for that quote. Both ini tial quotes received included a number of various errors, including Customer name, material quantities, units (ft vs lbs), and specified shipping locations. (Past experience example: We ordered some 145 gsm ¼”tape last year for an aero project. First contact with Hexcel SLC- April . First delivery (non-conforming material)- November. Final delivery of conforming product- Mid December.) Description: We ordered based on a Hexcel internal specificatio n but we tried to include additional requirements for defects, etc. Description: NASA requires a receipt for shipping as well as a shipping cost identified as a line item within a PO. Hexcel in cludes shipping costs in the per pound price of their material. This creates two issues for NASA procurement. (1) cost competiveness, and (2) no shipping costs to verify. During both procurement processes , NASA eventually caved, but not without spending weeks arguing over these points. Description: Hexcel quotes in ‘lbs’ but orders sent to manufacturing lists quantities in ‘ft’. Description: Both Hexcel and Cytec reps are very slow to respo nd; it has taken upwards of a week to get a question answered, modification made, or order acknowledged. CEUS utilized an accelerated building block approach through application of an equivalency test matrix. CMH17 requires a minimum of 8 coupons per test for equivalency. In this project , 18 coupons were submitted per test; 6 each from panels made at GRC, LaRC, and MSFC. We were not aware that the data generated from coupons prepared at separate locations could not be pooled without a separate statistical analysis (ANOVA). In the future, a minimum of 8 coupons per location is recommended. Executed the tooling procurement too early in the project. There was not sufficient design and analysis performed to support the tool. We bought a best effort surface that as the project progressed was not optimal for the structure. Tooling was long lead item - long lead schedule problem and then you need enough information to execute it. something for cure cycle development? Did we make improvements on cure cycles to improve properties (or did we get as far as property testing? Only saw a micrograph) learned low cost tooling is a possibility ? could only use some test fixtures could only test in certain buildings Testing in large facilities can become cost prohibitive for a p roject, so generally it is a benefit to test the smallest test article capable of achieving needed test objectives. The 1-Million pound machi ne at LaRC has an operating envelope of approximately 7' wide by 12' tall. Many of the design, analysis, materials, and manufacturing lessons learned were not evident in the coupon, element, or subcomponent testing and only started becoming evident when building the full-scale assembly. Issue / Benefit Time required to receive quote. Prepreg Specifications General Procurement Requirements Errors in material fabrication Non-responsiveness of material vendors Miniumum number of samples needed for statistics Phased tooling procurement too early / programmatic issue with accelerated project Cure Cycle Tooling Fixtures Facilities Panel test size restrictions Scale Up Date Category Analysis, Assembly, Design, Drawings, General, Handling, Instrumentation, Manufacturing, Testing Materials Materials Materials Materials Materials Manufacturing Manufacturing Manufacturing Test Test Test All LL ID

122 CEUS Historic Lessons Learned Review, Rev. 02 Mar 2015 Document Number Document Name Date LLIS # Lesson Learned Applicable? Plan n/a Final Report of the X-33 Liquid Hydrogen Tank Test Investigation Team May 2000 n/a The importance of a comprehensive building-block approach to ensure mission success yes Utlize a building-block approach. (Overall) n/a Final Report of the X-33 Liquid Hydrogen Tank Test Investigation Team May 2000 n/a The importance of communication. Communication is a two-way process. Technical information must be yes Do not compartmentalize between requirements, design, analysis, manufacturing, test, and communicated clearly. There must be openness to receive the information properly. inspection teams. (Overall) n/a Final Report of the X-33 Liquid Hydrogen Tank Test Investigation Team May 2000 n/a Failure modes must be addressed in depth. yes Consider/assess all potential failure modes to determine which ones are driving. I.e., do not make a priori assumption - all composites projects are different. (All analysis tasks) n/a Final Report of the X-33 Liquid Hydrogen Tank Test Investigation Team May 2000 n/a A risk management plan commensurate with the technical complexity must be used. TBD TBD n/a Final Report of the X-33 Liquid Hydrogen Tank Test Investigation Team May 2000 n/a Early expert reviews (at the design table) have the greatest effect on mission success. TBD TBD NASA TM 4322A NASA Preferred Reliability Practices (GD-ED-2210) Feb. 1999 689 Avoidance of the material selection considerations can result in a fiber composite part or component that yes Solicit and heed input from Manufacturing concerning material selection. (STA design) 1) does not satisfy design properties, 2) is unnecessarily difficult to fabricate, and 3) adds substantial program. Item 1 may lead to an expensive redesign. Item 2 can impact cost and program schedule. Item 3 can jeopardize the entire project. NASA TM 4322A NASA Preferred Reliability Practices (GD-ED-2205) Feb. 1999 682 Failure to use state-of-the-art design techniques, tooling, manufacturing techniques, and automated yes Set and encourage realistic expectations for mass savings. manufacturing and inspection techniques for composite materials could result in the choice of inappropriate materials, costly scrappage, and potential failures in use. Failure to use composites in appropriate applications could result in noncompetitive products with greater complexity, weight, or damage susceptibility. NASA TM 4322A NASA Preferred Reliability Practices (PD-ED-1217) Feb. 1999 669 Failure to adhere to proven and acceptable practices in the design, manufacture, and testing of structural yes Adhere to proven and acceptable practices. laminate composites will result in an unacceptably high rejection rate of fabricated composite parts. This rejection rate could be caused by unacceptable structural strength, delaminations, excessive (or inadequate) porosity, surface defects, inclusions, improper dimensions, or inadequate physical, thermal, or chemical properties as revealed in nondestructive testing or in destructive coupon testing. A high rejection rate could cause delays in the vehicle assembly, testing, and checkout process, and could possibly affect the launch schedule. Increased costs of manufacture would also result from a high rejection rate. NASA TM 4322A NASA Preferred Reliability Practices (PD-AP-1318) Feb. 1999 819 Not performing a complete and comprehensive stress analysis on the spacecraft structural components yes Perform a complete and comprehensive stress analysis. may lead to an inadequate design with unsafe or inefficient load paths. Without proper stress analysis, the objectives of minimum weight and a balanced design will not be met. Structural testing may also be misguided in that some components may be inadequately tested while others may be over-stressed. NASA TM 4322A NASA Preferred Reliability Practices (PT-TE-1422) Feb. 1999 765 Failure to detect cracks, flaws, and voids in aerospace materials through the proper use of ultrasonic yes Use appropriate NDE techniques. testing and other approved nondestructive evaluation methods could result in the use of weakened structures, unbonded propellants and insulation layers, and potential pressure vessel failures or burnthroughs due to increased propellant surface area, resulting in potential mission failure. NASA TM 4322A NASA Preferred Reliability Practices (GD-ED-2201) Feb. 1999 675 If the [fastener] guidelines recommended are not followed, greater numbers of fastener types, sizes, yes The requirements in NASA-STD-6016 (and NASA-STD-6008, which is levied in NASA-STD-6016) materials, and finishes may be specified or procured resulting in excessive cost. Mission performance may are applicable to CEUS as identified in the SRD and meet the intent of many of the guidelines be also degraded due to incompatibility of materials and finishes, or due to substandard hardware that in this preferred reliability practice. lacks sufficient screening and testing. NASA TM 4322A NASA Preferred Reliability Practices (PT-TE-1410) Feb. 1999 778 Noncompliance with outgassing requirements could result in degraded science data due to excessive yes Consider outgassing during material selection. contamination of an instrument, or in the complete failure of a space flight mission. Noncompliance with outgassing requirements could result in non-approval of materials for space flight use. If the non-approved materials are already assembled in a flight vehicle or scientific instrument, they may have to be removed and replaced by approved materials. n/a LLIS: "Space Charging of Composite Structures" March 2003 1330 Surface resistance measurements of a non-homogeneous composite material may not accurately TBD TBD characterize the material's susceptibility to surface charging. n/a LLIS: "Undocumented Process Steps And Process Drift Is Root Cause For Scrap Of High May 2012 6616 During the design development and manufacturing process development phase, establish configuration yes Some level of configuration control is necessary even for one-of-a-kind builds to ensure a Value Part"; Multi-Purpose Crew Vehicle (MPCV) controlled manufacturing processes. These manufacturing processes should establish the critical steps and quality product and avoid unexpected failures. provide detailed work instructions for the critical steps. Changes to the build process require a configuration controlled revision to these manufacturing processes. If significant time has elapsed or a new manufacturing crew is used, require them to hold a process walk thru of the manufacturing process to assure they understand the steps in the process, the equipment required, condition of the equipment, and materials required. The fit of the vacuum curing bag materials is a critical parameter for the curing of a composite structural part. The perforated release film and solid release film are susceptible to having unintended holes when the materials do not lay flat. The process step of cutting the materials into gore sections was not documented within the bagging process instructions. These steps need to be documented to assure that these steps performed consistently. IRIS Case Number: S- LLIS: "Cryogenic Tank Rupture/Close Call" Feb. 2011 5396 The main lesson is a lack of adequate independent reviews for the test plan, procedure, and operation. The no The CEUS is not pressurized hardware. The CEUS team is experienced in testing similar 2008-359-00002 NASA and contractor test team members did not seek an outside, independent review of a test that was structures. clearly beyond their expertise and experience. An independent review by other disciplines would have likely pointed out important aspects to consider and overcome the group-think and schedule pressure that were factors in many of the decisions made. This mishap would have been avoided had procedures/requirements been followed. When testing a Composite Pressure Vessel (CPV) tank above Maximum Expected Operating Pressure (MEOP), and with limited expertise/knowledge of CPV, an abundance of caution should have been employed. The involvement of Safety and Mission Assurance functions, various NASA discipline consultations, the performance of risk assessments, and the completion of variance documentation, would have triggered the right questions and analyses, thereby avoiding the mishap. Management review and insight for testing and risk analysis was inadequate. JPL Problem/Failure Heatshield Structure #1 (Flight Spare) Handling Drop Aug. 2008 1996 When ground handling flight hardware, even simple operations can fail without adequate procedures and yes Use trained personnel for ground handling operations. (STA manufacture and test) Report No. 13366 training. n/a LLIS: "To Bond or to Bolt, That is the Question [Export Version]"; OSTM/Jason 2 Nov. 2008 2038 1.Use of adhesive bonding for joining spacecraft fittings to structural components provides an opportunity yes TBD if CEUS skirts have bonded fittings. for elegant design solutions, but such bonds are subject to thermal stress and may be difficult to properly characterize and inspect. n/a LLIS: "To Bond or to Bolt, That is the Question [Export Version]"; OSTM/Jason 2 Nov. 2008 2038 2. This redacted (ITAR) lesson learned provides cautions regarding alternative methods of joining, such as yes Contact the office of the chief engineer at JPL to find out what the redacted lesson is. fasteners, that may provide an additional margin of safety. n/a LLIS: "To Bond or to Bolt, That is the Question [Export Version]"; OSTM/Jason 2 Nov. 2008 2038 3. This redacted (ITAR) lesson learned discusses the benefits of alternative methods of thermal stress yes Contact the office of the chief engineer at JPL to find out what the redacted lesson is. analysis for bonded joints. n/a LLIS: "To Bond or to Bolt, That is the Question [Export Version]"; OSTM/Jason 2 Nov. 2008 2038 4. This 4th redacted lesson learned provides guidelines for design and test of test coupons. yes Contact the office of the chief engineer at JPL to find out what the redacted lesson is. n/a LLIS: "To Bond or to Bolt, That is the Question [Export Version]"; OSTM/Jason 2 Nov. 2008 2038 5. This redacted (ITAR) lesson learned provides a material properties criterion for selecting a method for yes Contact the office of the chief engineer at JPL to find out what the redacted lesson is. joining mechanical components that does not overstress the hardware. NASA/TM-2011- Composite Crew Module: Primary Structure Nov. 2011 n/a LL-1: Small agile teams—emphasizing “in-line” work—working in parallel with, but outside the umbrella of no 217185; NESC-RP-06- a flight project can achieve significant results through rapid decision making. These teams can exploit 019 unforeseen opportunities and minimize consequence of risk with quick recovery to problems that may occur. NASA/TM-2011- Composite Crew Module: Primary Structure Nov. 2011 n/a LL-2: Engineering models predicted mass and structural response well. yes The tools being used by CEUS to create engineering models are the same or similar to what 217185; NESC-RP-06- was used for CCM. 019 NASA/TM-2011- Composite Crew Module: Primary Structure Nov. 2011 n/a LL-3: Engineering models did not always predict production challenges. no 217185; NESC-RP-06- 019 NASA/TM-2011- Composite Crew Module: Primary Structure Nov. 2011 n/a F-1: Many of the design, analysis, materials, and manufacturing lessons learned were not evident in the yes The CEUS project is aware that there can be scaling issues in manufacturing and analysis. The 217185; NESC-RP-06- coupon, element, or subcomponent testing and only became evident when building the full-scale project is going to build a full-scale test article for final verification of the design and 019 assembly. manufacturing. NASA/TM-2011- Composite Crew Module: Primary Structure Nov. 2011 n/a F-6: Small, but focused, independent technical reviews with subject matter experts were effective for yes TBD 217185; NESC-RP-06- challenging team assumptions and technical decisions resulting in increased innovation and a higher 019 motivated team. NASA/TM-2011- Composite Crew Module: Design Nov. 2011 n/a F-3: Structurally connecting the internal backbone structure to the bottom of the pressure shell produced no 217186; NESC-RP-06- the desired load-sharing effect to ultimately save significant structural weight. 019 NASA/TM-2011- Composite Crew Module: Design Nov. 2011 n/a F-4: Composite preform shapes are good structural design members for connecting intersecting composite yes Consider preform shapes for connecting the panels. (Joint task) 217186; NESC-RP-06- panels. 019 NASA/TM-2011- Composite Crew Module: Design Nov. 2011 n/a F-8: Attaching secondary structure to composite honeycomb panels was proven to be manageable and yes Consider this finding in trade studies as an advantage of using honeycomb. (STA design and 217186; NESC-RP-06- weight-efficient. Flight design) 019 NASA/TM-2011- Composite Crew Module: Materials and Processes Nov. 2011 n/a F-1: An alternative method to handle uncertainty in discontinuities is to use a single factor of safety TBD TBD 217187; NESC-RP-06- (consistent with that for acreage), but developed element level design allowables, as opposed to material 019 level allowables. This method handles the failure modes, variability, and uncertainty at a higher structural assembly level. NASA/TM-2011- Composite Crew Module: Materials and Processes Nov. 2011 n/a F-2: There is a gap between many of CCM's design allowables and true damaged material capability (i.e., TBD TBD 217187; NESC-RP-06- open hole compression versus compression after impact producing a 0.25-inch flaw). This gap represents a 019 weight saving opportunity, but requires developing both performance curves (e.g., strength versus defect size) and defect PoD curves/studies. NASA/TM-2011- Composite Crew Module: Materials and Processes Nov. 2011 n/a F-4: Minimum gage is partially driven by a derived permeability constraint. Decoupling leakage from TBD TBD 217187; NESC-RP-06- structure through use of a polymeric membrane would relieve this constraint, presenting a mass savings 019 opportunity in minimum gage reduction. NASA/TM-2011- Composite Crew Module: Materials and Processes Nov. 2011 n/a F-5: The 3D woven joints offered a robust joining solution for orthogonal joints. Specifically, their uncured yes Consider 3D woven joints. (Joint task) 217187; NESC-RP-06- flexibility accommodates large pre-cured assembly variation. Their fully interlocked weaves eliminated 019 delamination failure modes within the preform. Finally, they outperformed (strength/unit mass) conventional bonded joints for similar geometry. NASA/TM-2011- Composite Crew Module: Materials and Processes Nov. 2011 n/a F-6: Out-of-autoclave splicing/joining is a critical technology for enabling large launch structures without yes Perform analysis and testing of splicing/joining options. (Joint task) 217187; NESC-RP-06- the use of large infrastructure (i.e., large autoclaves). 019 NASA/TM-2011- Composite Crew Module: Analysis Nov. 2011 n/a F-5: Manufacturability, symmetry, and other derived requirements resulted in a CCM structure that does no No flight structure is ever perfectly optimized. All engineering designs are a compromise to 217188; NESC-RP-06- not have zero MS everywhere. These derived requirements can lead to higher MS thus higher mass than meet sometimes conflicting requirements. 019 expected. NASA/TM-2011- Composite Crew Module: Analysis Nov. 2011 n/a F-6: The CCM project included a final hydrostatic internal pressure test to failure. Evidence suggests that yes Do not ignore potential sources of nonlinearlity in higher-fidelity analyses. (STA and Flight 217188; NESC-RP-06- the failure occurred at a core splice between the upper shoulder core and the ceiling core. The failure, design) 019 exhibited as a facesheet disbond, is believed to be caused by through thickness tension/shear interaction as the shoulder radius opens under the pressure load. The core splice adhesive is stiffer than the surrounding honeycomb core, thereby drawing more load. Furthermore, it behaves in a brittle manner in tension, in contrast to the aluminum core. Therefore, the stress peaking in the splice material is exacerbated by the non-linear nature of the problem, causing a greater stiffness discrepancy between core and core splice. NASA/TM-2011- Composite Crew Module: Analysis Nov. 2011 n/a F-8: HyperSizer proved useful throughout the project, but excelled in three particular areas: 1. Early in the yes Hypersizer is being used for mass trades. (Analysis tasks) 217188; NESC-RP-06- design, HyperSizer enabled rapid trade studies between different architectural concepts such as sandwich 019 versus a variety of discrete stiffeners. 2. Once the architecture was selected, HyperSizer was used to determine the lightest weight layups to meet the required loads. While these layups were not always manufacturable, knowing the absolute lightest enabled an intelligent selection of layups throughout the structure. The manufactured layup was re-analyzed using HyperSizer to ensure that it also maintained positive MS. This process was formally incorporated into later versions of HyperSizer. 3. Once the design was released, HyperSizer was used in configuration management, ensuring that the effects of changes were tracked.

123 NASA/TM-2011- Composite Crew Module: Analysis Nov. 2011 n/a F-10: A coarse grid FEM (3-inch element size) was sufficient to release drawings for manufacturing of the yes Use models with appropriate and sufficient levels of fidelity for the different analysis tasks. 217188; NESC-RP-06- final design but the fine grid model (~1-inch element size) was constructed to better characterize the effect (All analysis tasks) 019 of ply drops and other details within the structure. The extra detail in this model allowed for more accurately representing layups, and was especially important for the development of the backbone cap. In preparation for static test, the detail of the fine grid model generated more accurate strain predictions and allowed much faster resolution of strain issues. NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-3: Including a full-scale manufacturing development unit as part of the fabrication process development no 217189; NESC-RP-06- prior to design release provided accurate scaled feedback with coupon/element testing and allowed 019 manufacturing lessons learned to influence the final design. NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-4: Using the final hardware cure tool to perform aluminum core heat forming process eliminated the TBD TBD 217189; NESC-RP-06- need for large limited use tooling. 019 NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-5: Alcore SHAPEGRID offered a robust core solution for high compound curvature regions. TBD TBD 217189; NESC-RP-06- 019 NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-6: The multi-step cure process (i.e., cure of inner skin, cure of core to inner skin bond, and cure of outer yes Consider a multi-step cure process for manufacturing of STA (STA manufacturing) 217189; NESC-RP-06- skin on the core as separate, sequential cures) proved to have numerous manufacturing advantages 019 NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-8. Male cure tools on the CCM inner surface were selected to increase the consolidation of the IML skin TBD TBD 217189; NESC-RP-06- to decrease porosity and increase the likelihood of low permeability skin. 019 NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-10: FiberSIM was an invaluable software tool that facilitated the complex manufacturing of the CCM yes Use of FiberSIM is planned. (Manufacturing analysis) 217189; NESC-RP-06- 019 NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-11: The need for repair was caused by a variety of sources including inadvertent impact damage, failed yes Take special care with tooling (tethers?) to avoid impact damage. Make sure procedures for 217189; NESC-RP-06- secondary cures, and secondary mechanical fastener hole misalignment. curing are followed correctly. Take special care with the placement of fastener holes; 019 designers make sure to accommodate existing location constraints, manufacturing make sure to follow drawings. (STA design and manufacturing) NASA/TM-2011- Composite Crew Module: Manufacturing Nov. 2011 n/a F-12: Pi-preforms offer a joint configuration that has manufacturing benefits that are less sensitive to fit-up yes Consider pi-preforms for joints. (Joint task) 217189; NESC-RP-06- clearance, and bond line thickness 019 NASA/TM-2011- Composite Crew Module: Test Nov. 2011 n/a F-1: Real-time monitoring of test results against analytical predictions was central to the success of the full- yes Generate pre-test predictions and have analysts at stress stations during test. (STA analysis 217190; NESC-RP-06- scale test program. This enabled the team to push the applied load limits progressively while minimizing and test) 019 the risk of catastrophic failure. NASA/TM-2011- Composite Crew Module: Test Nov. 2011 n/a F-2: Timely reconfiguration from test-to-test was enabled using low-stretch high-strength flexible straps in TBD TBD 217190; NESC-RP-06- combination with rollers to apply point loads. In addition to quick reconfiguration, the straps provided 019 accurate, predictable, and repeatable load orientation and simplified the mounting of the load cylinders. A 10–12-percent frictional loss was measured when the load was routed around a roller. NASA/TM-2011- Composite Crew Module: Test Nov. 2011 n/a F-4: While strain gages were the primary strain-measuring sensors for all full-scale tests, relatively newer yes Lobby for photogrammetry for testing of STA. (STA test) 217190; NESC-RP-06- techniques such as fiber optics and photogrammetry proved to be dependable and provided lineal 019 distribution of strain (fiber optics) and full-field strains (photogrammetry). In areas of high-strain gradients, these techniques were invaluable. NASA/TM-2011- Composite Crew Module: Test Nov. 2011 n/a F-5: To avoid possible interference between the painted speckled patterns and the post-test IR TBD TBD 217190; NESC-RP-06- thermography, removable (self-adhesive paper) speckled patterns were utilized in some regions. These did 019 not appear to affect photogrammetric measurements and the clean removal enabled improved post-test NDE inspections. NASA/TM-2011- Composite Crew Module: Test Nov. 2011 n/a F-6: Pressurizing the CCM to failure with water instead of air preserved the mode of failure and allowed the no 217190; NESC-RP-06- team to carry out an effective failure analysis. Preserving the structure also offered options for further 019 research studies. NASA/TM-2011- Composite Crew Module: Test Nov. 2011 n/a F-7: During all full-scale tests, the test team included members from all disciplines of the project. This TBD TBD 217190; NESC-RP-06- enabled real-time decision-making and contributed to the successful and timely completion of the full- 019 scale tests. NASA/TM-2011- Composite Crew Module: Nondestructive Evaluation Nov. 2011 n/a F-1: Woven joining technologies (i.e., Pi-preform and cruciform) offer robust solutions for orthogonal yes Be mindful of inspection considerations when choosing joint construction. (Joint trade) 217191; NESC-RP-06- joining. However, these technologies provide inspection challenges for bondline verification due to rough 019 surface texture and complex internal fiber geometry. A custom ultrasonic probe housing was required to successfully inspect the Pi-preform joint bondline on the lobed floor of the full-scale structure. Internal ppt from Mark Mark Kearney's Lessons Learned on Ares 1 Interstage Jan. 2011 n/a Composite testing is essential and should not be short changed TBD TBD Kearney Internal ppt from Mark Mark Kearney's Lessons Learned on Ares 1 Interstage Jan. 2011 n/a Testing should be completed in stages from testing/processing of basic features, through TBD TBD Kearney testing/processing of detailed design features, and through flight qualification testing. The schedule should be flexible since unknowns will need to be addressed. Internal ppt from Mark Mark Kearney's Lessons Learned on Ares 1 Interstage Jan. 2011 n/a Cutting testing will increase risk of component failures, increase costs of design changes, and drastically TBD TBD Kearney increases time required to resolve design issues. Internal ppt from Mark Composite Cryogenic Tank Demonstration Lessons Learned June 2013 n/a Compressed schedules: Risk is added when activities are planned in parallel. Time should be placed in TBD TBD Kearney for EV tech schedule for activities that feed into project decisions. luncheon Internal ppt from Mark Composite Cryogenic Tank Demonstration Lessons Learned June 2013 n/a Material Selection: Immaturity of material can add risk to project. Flight project [should] pick materials that yes Use conservative allowables for preliminary trade studies and design analyses. Use Kearney for EV tech have 20-50 years of experience. conservative allowables for design analyses until more accurate values obtained from testing luncheon specific to the project. If over-conservatism is a concern, select materials that are more well- characterized. Internal ppt from Mark Composite Cryogenic Tank Demonstration Lessons Learned June 2013 n/a Requirements: Make sure all parties have agreement as to the implementation of the structural criteria. yes Make sure required design factors of safety and test factors are understood. Make sure NDE Kearney for EV tech techniques and analysis implications (inspection limits for fracture analysis, etc.) are luncheon understood. Internal ppt from Mark Composite Cryogenic Tank Demonstration Lessons Learned June 2013 n/a Thermal Compensation: Strain gauges are sensitive to thermal effects; use consistent gauges throughout yes Perform sufficient test planning for thermal compensation of instrumentation if needed. Kearney for EV tech testing, or perform accurate thermal compensation. Small differences in composite lay-up between sample- luncheon size specimens and larger-scale articles affect thermal compensation. Internal doc from Rob Lessons Learned, SRB Composite Nose Cap Project, MSFC Strength Analysis Group Dec. 1999 n/a Material Characterization: The Strength Analysis Group should provide more insight to the project during yes Preliminary sizing studies are being performed for CEUS. Sensitivites to pristine vs. damaged Wingate the development of the material characterization plan... The Strength Analysis Group should provide more properties are being explored. The stress team is communicating with the materials and insight to the project during the development of the material characterization plan... requires the Strength testing groups concerning the damage tolerance plan and material propertied needed. Analysis Group to start doing preliminary analyses at project inception. These analyses should include all major loads, environments, and structural configurations to help determine which properties are needed. The preliminary analyses should also be used to determine which analytical techniques, finite element modeling strategies, etc. are going to be used as that may also determine which material properties are needed. Internal doc from Rob Lessons Learned, SRB Composite Nose Cap Project, MSFC Strength Analysis Group Dec. 1999 n/a Preliminary Analysis: The Strength Analysis Group should begin preliminary analyses at project inception yes Preliminary sizing studies are being performed for CEUS. Sensitivites to pristine vs. damaged Wingate using educated assumptions for material properties if necessary. By the time A-basis strength allowables properties are being explored. The stress team is communicating with the materials and were available, the CNC project was well underway, with fabrication of the second prototype (P2) testing groups concerning the damage tolerance plan and material propertied needed. complete, and production of the third (final) prototype and first qualification unit scheduled to begin soon. However, the A-basis allowables resulted in significant negative margins of safety in several areas of the composite shell, requiring redesign should the project have not been cancelled. Earlier analyses had used preliminary average properties, which were much higher than the A-basis properties. It could be argued that the schedule was too aggressive and should have allowed more time between material characterization and production of the qualification units. However, the Strength Analysis Group should also have worked with the Materials engineers to make educated assumptions for material properties to facilitate more accurate analyses early in the project. These assumptions could include use of knockdown factors on average properties, use of properties for similar materials used on other projects, etc. Internal doc from Rob Lessons Learned, SRB Composite Nose Cap Project, MSFC Strength Analysis Group Dec. 1999 n/a Damage Tolerance/Fracture Control: Damage tolerance of composite structures should be considered from yes Preliminary sizing studies are being performed for CEUS. Sensitivites to pristine vs. damaged Wingate project inception. Also, the MSFC Fracture Control Board should be consulted early in a project to gain properties are being explored. The stress team is communicating with the materials and their insight, identify any show-stoppers, and get them “on-board.” A fracture control plan should be testing groups concerning the damage tolerance plan and material propertied needed. required. Damage tolerance was not addressed until well into the project and by the time the project was cancelled, a damage tolerance plan still had not been fully developed. It is not known why it was Fracture Control Board consultation? TBD overlooked initially, but an informal meeting with some members of the MSFC Fracture Control Board raised the issue. The Strength Analysis Group, with support from the Materials organization, should develop a damage tolerance plan either separately or as part of the materials characterization plan or fracture control plan. This planning should begin at project inception, as it may be costly due to the testing and inspection required. It may also influence early material trade studies. In order to prevent such an oversight again, it is recommended that the MSFC Fracture Control Board be consulted early in a project to gain their insight, identify any show-stoppers, and get them “on-board.” The Board should have been involved anyway as part of the fracture control review process, but no fracture control plan was required for the CNC. It is recommended that any future project like the CNC require a fracture control plan as well.

Internal doc from Rob Lessons Learned, SRB Composite Nose Cap Project, MSFC Strength Analysis Group Dec. 1999 n/a Structural Verification: A Structural Verification Plan should be written, signed by the Leader of the yes TBD Wingate Strength Analysis Group. Internal doc from Rob Lessons Learned from the SRB Composite Nose Cap Project Jan. 2000 n/a At the initiation of the project, a review of the preliminary requirements was conducted by the Composite yes A Structural Requirements Review (SRR) was conducted on Feb. 13, 2015. Wingate Nose Cap team. Significant changes to the testing and analysis requirements were introduced throughout the design process. Based on the continuous change in testing and analysis requirements, it was identified that a formal Preliminary Requirements Review (PRR) should have been conducted. Internal doc from Rob Lessons Learned from the SRB Composite Nose Cap Project Jan. 2000 n/a The Preliminary Design Review was conducted in March, 1998. Though the review produced a good review yes TBD Wingate of the material provided, the primary participants in the review we on the design team. Subsequent to the review, other representatives from MSFC and USBI became involved in the project and identified significant impacts in the areas of testing and analysis. The introduction of the additional testing and analysis provided for significant cost and schedule impacts to the project. The identification of these impacts late in the project could have been avoided by a more thorough review team at the Preliminary Design Review. Internal doc from Rob Lessons Learned from the SRB Composite Nose Cap Project Jan. 2000 n/a The Material Characterization Plan outlined all material level testing that would be performed in the yes TBD Wingate material characterization phase of the project. Though the testing was identified in the plan, the actual testing conditions and approach were not determined until much later in the project. In several cases, testing was delayed until the test approach could be agreed upon by the technical community. Moisture conditioning of test samples for mechanical and thermal testing is one area where it was identified that the conditioning used was significantly more conservative that the actual conditioning observed during beach exposure testing. It was also identified that all samples (including dry samples) should be weighed prior to testing to determine any moisture gain during storage. Internal doc from Rob Lessons Learned from the SRB Composite Nose Cap Project Jan. 2000 n/a Design, development, test and evaluation (DDT&E) projects historically have cost, schedule, and weight TBD TBD Wingate growth throughout the project. These impacts can be minimized by taking the following approaches: (1) Lay the project out in Phases (Material Characterization, Preliminary Design/Analysis, Final Design/Analysis and Qualification). This approach would allow for periodic funding and schedule reviews prior to continuing to the next phase. (2)Partnering of projects between the contractor and NASA can provide for benefits to both parties. Conduct a trade study to determine if the project fits within the partnering guidelines as well as the capabilities of parties to be involved. Clearly document deliverables and required delivery dates for data that must be shared between the parties.

124 Internal doc from Rob Lessons Learned from the SRB Composite Nose Cap Project Jan. 2000 n/a Clear management direction and support for the project should be demonstrated prior to initiating the TBD TBD Wingate project. Once the project is initiated, priorities should be placed to assure all schedule critical items are completed on time. Internal doc from Rob Lessons Learned from the SRB Composite Nose Cap Project Jan. 2000 n/a Changes that provide for cost or schedule impact should be addressed immediately when identified. TBD TBD Wingate Internal pdf Composite Interstage Structural Concepts for Heavy Lift Launch Vehicles; Exploration Sep. 2011 n/a Ensure that all penetrations and other details from the CAD model and the previous analysis model TBD TBD Technology Development Program, Advanced Composites Technology (ACT) correspond. Structural Concepts; Lightweight Spacecraft Structures & Materials (LSSM) Project Internal pdf Composite Interstage Structural Concepts for Heavy Lift Launch Vehicles; Exploration Sep. 2011 n/a Determine the areas that will be designated as discontinuities early in the analysis process. TBD TBD Technology Development Program, Advanced Composites Technology (ACT) Structural Concepts; Lightweight Spacecraft Structures & Materials (LSSM) Project Internal pdf Composite Interstage Structural Concepts for Heavy Lift Launch Vehicles; Exploration Sep. 2011 n/a The ring frames and longerons being modeled in the proper location had a considerable impact on the TBD TBD Technology Development Program, Advanced Composites Technology (ACT) stress field around the penetrations. Structural Concepts; Lightweight Spacecraft Structures & Materials (LSSM) Project Internal pdf Composite Interstage Structural Concepts for Heavy Lift Launch Vehicles; Exploration Sep. 2011 n/a Any model used for a concept down select, even when using beam or bar elements, should employ offsets TBD TBD Technology Development Program, Advanced Composites Technology (ACT) to obtain a better stress field. Structural Concepts; Lightweight Spacecraft Structures & Materials (LSSM) Project NASA-CR-4620 Composite Chronicles: A Study of the Lessons Learned in the Development, Nov. 1994 n/a The use of a basic laminate family containing 0/90/+45 plies with a minimum of 10% of the plies in each TBD TBD Production, and Service of Composite Structures direction is well suited to most applications, generally assures fiber dominated laminate properties, and simplifies layup and inspection. NASA-CR-4620 Composite Chronicles: A Study of the Lessons Learned in the Development, Nov. 1994 n/a The number of mechanical joints should be minimized by utilizing large cocured or cobonded TBD TBD Production, and Service of Composite Structures subassemblies. Mechanical joints should be restricted to attachment of metal fittings and situations where assembly or access is impractical using alternative approaches. NASA-CR-4620 Composite Chronicles: A Study of the Lessons Learned in the Development, Nov. 1994 n/a Cocuring and cobonding are preferred over secondary bonding. Secondary bonding requires near perfect TBD TBD Production, and Service of Composite Structures interface fit-up. NASA-CR-4620 Composite Chronicles: A Study of the Lessons Learned in the Development, Nov. 1994 n/a Mechanically fastened joints require close tolerance fit-up. Liquid or structural shimming is usually TBD TBD Production, and Service of Composite Structures required to assure a good fit and to avoid damage to the composite parts during assembly. NASA-CR-4620 Composite Chronicles: A Study of the Lessons Learned in the Development, Nov. 1994 n/a Dimensional tolerances are more critical in composites than in metals. Dimensional control of mating TBD TBD Production, and Service of Composite Structures surfaces can reduce assembly costs and avoid damage to parts during assembly. NASA-CR-4620 Composite Chronicles: A Study of the Lessons Learned in the Development, Nov. 1994 n/a Large, cocured assemblies reduce part count and assembly costs. If the cocured assembly requires overly TBD TBD Production, and Service of Composite Structures complex tooling, however, the potential cost savings from low part count can be easily negated. Producibility must be a key consideration in the design. Designing for producibility is generally more cost effective than optimization for weight savings. Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Design structure to be testable. Much easier and less expensive to do qual testing at room temperature: (1) yes TBD Boeing/Don Barnes Payload Attach Structures Requires knowledge of time consistent loads and thermal profiles to identify worst case; (2) Test at RT amd use Environmental Correction Factor; (3) Thermally condition interfaces only if significant temperature differences are present, such as at cryogenic joints. Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Perform material testing as early as possible, have a good set of allowables in time for PDR. Perform testing yes TBD Boeing/Don Barnes Payload Attach Structures early to prevent downstream requirements changes. Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a For producibility, IM7/8552-1 (vs. IM7/8552) is specially formulated to run through machines better and TBD TBD Boeing/Don Barnes Payload Attach Structures have longer out-time. Honeycomb core 5056P has better properties and better corrosion resistance than 5052. Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Develop repair procedures early. TBD TBD Boeing/Don Barnes Payload Attach Structures Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Don’t forget Secondary Structure attachment allowable testing - allowables determined analytically TBD TBD Boeing/Don Barnes Payload Attach Structures conservative compared to tested allowables Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Identify and freeze lift point (transportation and handling) locations early to assist with GSE design. yes TBD Boeing/Don Barnes Payload Attach Structures Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Use segmented rings for attaching to composite shell [if shell is a one-piece cylinder]. Consider placing TBD TBD Boeing/Don Barnes Payload Attach Structures rings against IML instead of OML for load distribution -- may result in better load path and use of smooth faying surface [depending on joint configuration and tooling surface]. Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a For joint design: use squared-off designs - allows double lap shear on core and adjoining structure, allows TBD TBD Boeing/Don Barnes Payload Attach Structures for one piece closeout, promotes concentricity in load path, can avoid ply drops on chamfered core, can avoid sharp points in core (which can be easily damaged, driving up rework). Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Core mismatch criteria needs to be defined early as it potentially affects required facesheet thickness. TBD TBD Boeing/Don Barnes Payload Attach Structures Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Use high density core rather than potting at bolted joint locations. Potting involves more touch labor, adds TBD TBD Boeing/Don Barnes Payload Attach Structures cost, and is rigid and does not conform to the tool surface. Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Do not sculpt core for pad-ups. TBD TBD Boeing/Don Barnes Payload Attach Structures Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Where separate doublers end up being close together or overlapping, combine doublers into a larger yes TBD Boeing/Don Barnes Payload Attach Structures reinforcement. Applicable to both openings and secondary structure attachment. Also applies to areas where higher density core is needed. Combining doublers reduces the number of plies that have to be cut and kitted, and also eases the creation of design and stress models.

Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Define Access Requirements Now. Scope out GSE interfaces (and stay-out zones) now, based on yes TBD Boeing/Don Barnes Payload Attach Structures reasonable assumptions and experience if there are no hard requirements. Internal ppt from Boeing Lessons Learned: Delta IV Centerbody and Interstage, Sea Launch Fairings and Oct. 2007 n/a Pay attention to the mass ratios of the stages – they help to identify where you get the most “bang for the yes TBD Boeing/Don Barnes Payload Attach Structures buck” in terms of weight savings vs adding mass for producibility. Possibly more cost-effective to add the weight to the structure (producibility) to pay dividends later (extensibility, reusability).

125 APPENDIX B––MATERIALS

The test matrices in tables 27–29 were being implemented in support of the Digimat modeling effort. Fiber had been procured and resin panel procurement was in process at the time of this publication. Panels from the equivalency test program were made available to this effort, but no test coupons have been tested at the time of this publication.

Table 27. Fiber (IM7) test matrix.

No. of Batches × No. of Panels × No. of Specimens Test Temperature/Moisture Condition Layup Test Type and Direction Property CTD RTD ETW

Single-filament ASTM D3379 (withdrawn)–axial, tension E11, +S1 15 15 15

Single-filament Tensile recoil test –S1 15 15 15

Single-filament Torsion pendulum test GLT 15 15 15 Total number of tests: 135

Table 28. Matrix (8552-1 neat resin) test matrix.

No. of Batches × No. of Panels × No. of Specimens Test Temperature/Moisture Condition Layup/Material Test Type and Direction Property CTD RTD ETW Matrix, resin ASTM D638, tension E, +S, nu 1×3×5 1×3×5 1×3×5 Matrix, resin ASTM D695, compression E, –S, nu 1×3×5 1×3×5 1×3×5 Matrix, resin ASTM D5379, losipescu shear G, S12 1×3×5 1×3×5 1×3×5

Total number of tests: 135

Table 29. IM7/8552-1 lamina/laminate test matrix.

No. of Batches × No. of Panels × No. of Specimen Test Temperature/Moisture Condition Layup Test Type and Direction Property CTD RTD ETW

[0]6 ASTM D3039, tension E11, +S1, V12 1×3×5 1×3×5 1×3×5

[0]14 ASTM D6641, compression E11c, –S1 1×3×5 1×3×5 1×3×5

[90]14 ASTM D3039, tension E22, +S2, V12 1×3×5 1×3×5 1×3×5

[90]14 ASTM D6641, compression E22c, –S2 1×3×5 1×3×5 1×3×5

[45/–45]3s ASTM D2344, shear G12, S12 1×3×5 1×3×5 1×3×5

[45/0/–45/90]s ASTM D3039, tension Ex or y, Fx or y 1×3×5 1×3×5 1×3×5

[45/0/–45/90]s ASTM D6641, compression Ex or y, Fx or y 1×3×5 1×3×5 1×3×5

[45/0/–45/90]s ASTM D3039, tension Ex or y, Fx or y 1×3×5 1×3×5 1×3×5

[45/0/–45/90]s ASTM D6641, compression Ex or y, Fx or y 1×3×5 1×3×5 1×3×5

Total number of tests: 405 126 B.1 8552 IM7 Unidirectional Prepreg Reports

Hexcel 8552 IM7 Unidirectional Prepreg 190 gsm and 35% RC Statistical Analaysis Report for Combined NASA Samples and the Equivalent Statistical Analysis Report are found in this appendix.

127

Report No: NCP-RP-2016-001 NC Report Date: April 11th, 2016

Hexcel 8552 IM7 Unidirectional Prepreg 190 gsm & 35%RC Statistical Analysis Report for Combined NASA Samples (Glenn, Langley & Marshall)

Test Report Number: NCP-RP-2016-001 NC

Report Date: April 11, 2016

Elizabeth Clarkson National Institute for Aviation Research Wichita State University Wichita, KS 67260-0093

Testing Facility: Composites Laboratory National Institute for Aviation Research Wichita State University 1845 N. Fairmount Wichita, KS 67260-0093

Test Panel Fabrication Facility: NASA Glenn Research Center NASA Langley Research Center NASA Marshall Space Flight Center

128 129 April 11, 2016 NCP-RP-2016-001 NC

Table of Contents

1. Introduction ...... 6 1.1 Symbols and Abbreviations ...... 6 2. Background ...... 8 2.1 Results Codes ...... 8 2.2 Equivalency Computations ...... 8 2.2.1 Hypothesis Testing...... 8 2.2.2 Type I and Type II errors ...... 9 2.2.3 Cumulative Error Probability ...... 9 2.2.4 Strength and Modulus Tests...... 10 2.2.5 Modified Coefficient of Variation ...... 12 3. Equivalency Test Results ...... 14 4. Individual Test Results ...... 20 4.1 Longitudinal (0°) Tension (LT) ...... 20 4.2 Quasi Isotropic (“25/50/25”) Unnotched Tension (UNT1)...... 26 4.3 Quasi Isotropic “25/50/25” Open-Hole Tension 1 (OHT1) ...... 31 4.4 Laminate Short-Beam Strength (LSBS) ...... 33 4.5 Quasi Isotropic (“25/50/25”) Unnotched Compression 1 (UNC1) ...... 35 4.6 Quasi Isotropic “25/50/25” Open-Hole Compression 1 (OHC1) ...... 39 5. Summary of Results ...... 42 5.1 Failures ...... 43 5.2 Pass Rate ...... 43 5.3 Probability of Failures ...... 43 6. Generic Basis Values Assessment ...... 45 7. References ...... 49

Page 3 of 49

130 April 11, 2016 NCP-RP-2016-001 NC

Figures Figure 2-1 Type I and Type II errors ...... 9 Figure 3-1 Summary of Combined Strength means and minimums compared to their respective Equivalence limits ...... 19 Figure 3-2 Summary of Combined Modulus means and Equivalence limits ...... 19 Figure 4-1 Longitudinal Tension Values, Averages and Acceptance limits for CTD Condition...... 24 Figure 4-2 Longitudinal Tension Values, Averages and Acceptance limits for RTD Condition...... 24 Figure 4-3 Longitudinal Tension Values, Averages and Acceptance limits for ETW Condition...... 25 Figure 4-4 UNT1 Values, Averages and Acceptance limits for CTD Condition ...... 29 Figure 4-5 UNT1 Values, Averages and Acceptance limits for RTD Condition ...... 30 Figure 4-6 UNT1 Values, Averages and Acceptance limits for ETW Condition ...... 30 Figure 4-7 OHT1 Values, Averages and Acceptance limits ...... 32 Figure 4-8 Laminate Short-Beam Strength means, minimums and Acceptance limits for NASA Labs ...... 34 Figure 4-9 UNC1 means, minimums and Acceptance limits for RTD Condition ...... 37 Figure 4-10 UNC1 means, minimums and Acceptance limits for ETW Condition ...... 38 Figure 4-11 OHC1 means, minimums and Acceptance limits ...... 41 Figure 5-1 Probability of Number of Failures ...... 44 Figure 6-1 NASA OHC1 results with NCAMP Generic Criteria and Basis values ...... 46 Figure 6-2 NASA LT & OHT1 CTD results with NCAMP Generic Criteria and Basis values ...... 47 Figure 6-3 NASA LT & OHT1 RTD results with NCAMP Generic Criteria and Basis values ...... 47 Figure 6-4 NASA LT & OHT1 ETW results with NCAMP Generic Criteria and Basis values ...... 48

Page 4 of 49

131 April 11, 2016 NCP-RP-2016-001 NC

Tables Table 0-1 Test Property Abbreviations ...... 6 Table 0-2 Environmental Conditions Abbreviations ...... 7 Table 2-1 One-sided tolerance factors for limits on sample mean values ...... 11 Table 2-2 One-sided tolerance factors for limits on sample minimum values ...... 12 Table 3-1 Summary of Equivalency Test Results for NASA Combined Locations...... 15 Table 3-2 Summary of Equivalency Test Results for NASA Glenn ...... 16 Table 3-3 Summary of Equivalency Test Results for NASA Langley ...... 17 Table 3-4 Summary of Equivalency Test Results for NASA Marshall ...... 18 Table 3-5 "% Failed" Results Scale ...... 18 Table 4-1 Longitudinal Tension Strength Results for CTD Condition...... 20 Table 4-2 Longitudinal Tension Strength Results for RTD Condition...... 21 Table 4-3 Longitudinal Tension Strength Results for ETW Condition ...... 21 Table 4-4 Longitudinal Tension Modulus Results for CTD Condition ...... 21 Table 4-5 Longitudinal Tension Modulus Results for RTD Condition ...... 22 Table 4-6 Longitudinal Tension Modulus Results for ETW Condition ...... 22 Table 4-7 UNT1 Strength Results for CTD Condition ...... 26 Table 4-8 UNT1 Strength Results for RTD Condition ...... 27 Table 4-9 UNT1 Strength Results for ETW Condition ...... 27 Table 4-10 UNT1 Modulus Results for CTD Condition ...... 28 Table 4-11 UNT1 Modulus Results for RTD Condition ...... 28 Table 4-12 UNT1 Modulus Results for ETW Condition ...... 29 Table 4-13 OHT1 Strength Results for CTD ...... 31 Table 4-14 OHT1 Strength Results for RTD ...... 31 Table 4-15 OHT1 Strength Results for ETW ...... 32 Table 4-16 Laminate Short-Beam Strength Results for RTD ...... 33 Table 4-17 Laminate Short-Beam Strength Results for ETW ...... 33 Table 4-18 UNC1 Strength Results for RTD ...... 35 Table 4-19 UNC1 Strength Results for ETW ...... 35 Table 4-20 UNC1 Modulus Results for RTD ...... 36 Table 4-21 UNC1 Modulus Results for ETW ...... 36 Table 4-22 OHC1 Strength Results For RTD Condition ...... 39 Table 4-23 OHC1 Strength Results For ETW Condition ...... 39 Table 6-1 OHC1 Generic Basis Values Acceptance Criteria and NASA test results ...... 45 Table 6-2 LT Generic Basis Values Acceptance Criteria and NASA test results ...... 46 Table 6-3 OHT1 Generic Basis Values Acceptance Criteria and NASA test results ...... 46

Page 5 of 49

132 April 11, 2016 NCP-RP-2016-001 NC

1. Introduction

This report contains the equivalency test results for NASA Hexcel 8552 IM7 unidirectional material produced panels compared to the original qualification panels of the same material. The lamina and laminate material property data have been generated with a single batch of material. Panels were produced at the Glenn, Marshall and Langley research centers.

The tests on the equivalency specimens were performed at the National Institute for Aviation Research (NIAR) in Wichita, Kansas. The comparisons were performed according to CMH-17 Rev G section 8.4.1. The modified coefficient of variation (Mod CV) comparison tests were done in accordance with section 8.4.4 of CMH-17 Rev G.

The material property data for the qualification panels is published in Property Data Report CAM-RP-2009-015 Rev A. The material property data for the Integrated Composites, Inc equivalence panels is published in NCAMP Test Report CAM-RP-2012- 015 N/C. Engineering basis values computed using CMH-17 methodologies were reported in NCAMP Report NCP-RP-2009-028 N/C which details the standards and methodology used for computing basis values as well as providing the B-basis values and A- and B- estimates computed from the test results for the original qualification panels. An alternative methodology, generic basis values, was developed for this material using data from multiple locations and was published in NCP-RP-2013-015 N/C.

The NCAMP shared material property database contains material property data of common usefulness to a wide range of aerospace projects. However, the data may not fulfill all the needs of a project. Specific properties, environments, laminate architecture, and loading situations that individual projects need may require additional testing.

The applicability and accuracy of this material property data, material allowables, and specifications must be evaluated on case-by-case basis by aircraft companies and certifying agencies. NIAR assumes no liability whatsoever, expressed or implied, related to the use of the material property data, material allowables and specifications.

1.1 Symbols and Abbreviations

Test Property Abbreviation Longitudinal Tension LT Unnotched Tension 1 UNT1 Unnotched Compression 1 UNC1 Laminate Short-Beam Strength LSBS Open-Hole Tension 1 OHT1 Open-Hole Compression 1 OHC1 Table 0-1 Test Property Abbreviations

Page 6 of 49

133 April 11, 2016 NCP-RP-2016-001 NC

Environmental Condition Temperature Abbreviation Cold Temperature Dry −65º F CTD Room Temperature Dry 70º F RTD Elevated Temperature Wet 250º F ETW Table 0-2 Environmental Conditions Abbreviations

Tests with a number immediately after the abbreviation indicate the lay-up:

1 refers to a 25/50/25 layup. This is also referred to as "Quasi-Isotropic" 2 refers to a 10/80/10 layup. This is also referred to as “Soft” 3 refers to a 40/20/40 layup. This is also referred to as “Hard”

EX: OHT1 is an Open-Hole Tension test with a 25/50/25 layup.

Page 7 of 49

134 April 11, 2016 NCP-RP-2016-001 NC

2. Background

Equivalence tests are performed in accordance with section 8.4.1 of CMH-17 Rev G and section 6.1 of DOT/FAA/AR-03/19, “Material Qualification and Equivalency for Polymer Matrix Composite Material Systems: Updated Procedure.”

2.1 Results Codes

Pass indicates that the test results are equivalent for that environment under both computational methods.

Fail indicates that the test results are NOT equivalent under both computational methods.

Pass with mod CV indicates the test results are equivalent under the assumption of the modified CV method that the coefficient of variation is at least 6 but the test results fail without the use of the modified CV method.

2.2 Equivalency Computations

Equivalency tests are performed to determine if the differences between test results can be reasonably explained as due to the expected random variation of the material and testing processes. If so, we can conclude the two sets of tests are from ‘equivalent’ materials.

2.2.1 Hypothesis Testing

This comparison is performed using the statistical methodology of hypothesis testing. Two mutually exclusive hypotheses are set up, termed the null (H0) and the alternative (H1). The null hypothesis is assumed true and must contain the equality. For equivalency testing, they are set up as follows, with M1 and M2 representing the two materials being compared:

HM:  M 01 2 HM11:  M 2

Samples are taken of each material and tested according to the plan. A test statistic is computed using the data from the sample tests. The probability of the actual test result is computed under the assumption of the null hypothesis. If that result is sufficiently unlikely then the null is rejected and the alternative hypothesis is accepted as true. If not, then the null hypothesis is retained as plausible.

Page 8 of 49

135 April 11, 2016 NCP-RP-2016-001 NC

2.2.2 Type I and Type II errors

Materials Materials are not are equal equal

Conclude Correct Type II materials Decision error are equal

Conclude materials Type I Correct are not error Decision equal Figure 2-1 Type I and Type II errors

As illustrated in Figure 2-1, there are four possible outcomes: two correct conclusions and two erroneous conclusions. The two wrong conclusions are termed type I and type II errors to distinguish them. The probability of making a type I error is specified using a parameter called alpha (α), while the type II error is not easily computed or controlled. The term ‘sufficiently unlikely’ in the previous paragraph means, in more precise terminology, the probability of the computed test statistic under the assumption of the null hypothesis is less than α.

For equivalency testing of composite materials, α is set at 0.05 which corresponds to a confidence level of 95%. This means that if we reject the null and say the two materials are not equivalent with respect to a particular test, the probability that this is a correct decision is no less than 95%.

2.2.3 Cumulative Error Probability

Each characteristic (such as Longitudinal Tension strength or In-Plane Shear modulus) is tested separately. While the probability of a Type I error is the same for all tests, since many different tests are performed on a single material, each with a 5% probability of a type I error, the probability of having one or more failures in a series of tests can be much higher.

If we assume the two materials are identical, with two tests the probability of a type I error for the two tests combined is 1 − .952 = .0975. For four tests, it rises to 1 − .954 = 25 .1855. For 25 tests, the probability of a type I error on 1 or more tests is 1 − .95 = .7226. With a high probability of one or more equivalence test failures due to random

Page 9 of 49

136 April 11, 2016 NCP-RP-2016-001 NC chance alone, a few failed tests should be allowed and equivalence may still be presumed provided that the failures are not severe.

2.2.4 Strength and Modulus Tests

For strength test values, we are primarily concerned only if the equivalence sample shows lower strength values than the original qualification material. This is referred to as a ‘one-sided’ hypothesis test. Higher values are not considered a problem, though they may indicate a difference between the two materials. The equivalence sample mean and sample minimum values are compared against the minimum expected values for those statistics, which are computed from the qualification test result

The expected values are computed using the values listed in Table 2-1 and Table 2-2 according to the following formulas:

table 2.1 The mean must exceed Xkn S where X and S are, respectively, the mean and the standard deviation of the qualification sample.

table 2.2 The sample minimum must exceed Xkn S where X and S are, respectively, the mean and the standard deviation of the qualification sample.

If either the mean or the minimum falls below the expected minimum, the sample is considered to have failed equivalency for that characteristic and the null hypothesis is rejected. The probability of failing either the mean or the minimum test (the α level) is set at 5%.

For Modulus values, failure occurs if the equivalence sample mean is either too high or too low compared to the qualification mean. This is referred to as a ‘two-sided’ hypothesis test. A standard two-sample two-tailed t-test is used to determine if the mean from the equivalency sample is sufficiently far from the qualification sample mean to reject the null hypothesis. The probability of a type I error is set at 5%.

These tests are performed with the HYTEQ spreadsheet, which was designed to test equivalency between two materials in accordance with the requirements of CMH-17 Rev G section 8.4.1: Tests for determining equivalency between an existing database and a new dataset for the same material. Details about the methods used are documented in the references listed in Section 0.

Page 10 of 49

137 April 11, 2016 NCP-RP-2016-001 NC

One-sided tolerance factors for limits on sample mean values a n 0.25 0.1 0.05 0.025 0.01 0.005 0.0025 0.001 0.0005 2 0.6266 1.0539 1.3076 1.5266 1.7804 1.9528 2.1123 2.3076 2.4457 3 0.5421 0.8836 1.0868 1.2626 1.4666 1.6054 1.7341 1.8919 2.0035 4 0.4818 0.7744 0.9486 1.0995 1.2747 1.3941 1.5049 1.6408 1.7371 5 0.4382 0.6978 0.8525 0.9866 1.1425 1.2488 1.3475 1.4687 1.5546 6 0.4048 0.6403 0.7808 0.9026 1.0443 1.1411 1.2309 1.3413 1.4196 7 0.3782 0.5951 0.7246 0.8369 0.9678 1.0571 1.1401 1.2422 1.3145 8 0.3563 0.5583 0.6790 0.7838 0.9059 0.9893 1.0668 1.1622 1.2298 9 0.3379 0.5276 0.6411 0.7396 0.8545 0.9330 1.0061 1.0959 1.1596 10 0.3221 0.5016 0.6089 0.7022 0.8110 0.8854 0.9546 1.0397 1.1002 11 0.3084 0.4790 0.5811 0.6699 0.7735 0.8444 0.9103 0.9914 1.0490 12 0.2964 0.4593 0.5569 0.6417 0.7408 0.8086 0.8717 0.9493 1.0044 13 0.2856 0.4418 0.5354 0.6168 0.7119 0.7770 0.8376 0.9121 0.9651 14 0.2760 0.4262 0.5162 0.5946 0.6861 0.7488 0.8072 0.8790 0.9300 15 0.2673 0.4121 0.4990 0.5746 0.6630 0.7235 0.7798 0.8492 0.8985 16 0.2594 0.3994 0.4834 0.5565 0.6420 0.7006 0.7551 0.8223 0.8700 17 0.2522 0.3878 0.4692 0.5400 0.6230 0.6797 0.7326 0.7977 0.8440 18 0.2455 0.3771 0.4561 0.5250 0.6055 0.6606 0.7120 0.7753 0.8202 19 0.2394 0.3673 0.4441 0.5111 0.5894 0.6431 0.6930 0.7546 0.7984 20 0.2337 0.3582 0.4330 0.4982 0.5745 0.6268 0.6755 0.7355 0.7782 21 0.2284 0.3498 0.4227 0.4863 0.5607 0.6117 0.6593 0.7178 0.7594 22 0.2235 0.3419 0.4131 0.4752 0.5479 0.5977 0.6441 0.7013 0.7420 23 0.2188 0.3345 0.4041 0.4648 0.5359 0.5846 0.6300 0.6859 0.7257 24 0.2145 0.3276 0.3957 0.4551 0.5246 0.5723 0.6167 0.6715 0.7104 25 0.2104 0.3211 0.3878 0.4459 0.5141 0.5608 0.6043 0.6579 0.6960 26 0.2065 0.3150 0.3803 0.4373 0.5041 0.5499 0.5926 0.6451 0.6825 27 0.2028 0.3092 0.3733 0.4292 0.4947 0.5396 0.5815 0.6331 0.6698 28 0.1994 0.3038 0.3666 0.4215 0.4858 0.5299 0.5710 0.6217 0.6577 29 0.1961 0.2986 0.3603 0.4142 0.4774 0.5207 0.5611 0.6109 0.6463 30 0.1929 0.2936 0.3543 0.4073 0.4694 0.5120 0.5517 0.6006 0.6354 Table 2-1 One-sided tolerance factors for limits on sample mean values

Page 11 of 49

138 April 11, 2016 NCP-RP-2016-001 NC

One-sided tolerance factors for limits on sample minimum values a n 0.25 0.1 0.05 0.025 0.01 0.005 0.0025 0.001 0.0005 2 1.2887 1.8167 2.1385 2.4208 2.7526 2.9805 3.1930 3.4549 3.6412 3 1.5407 2.0249 2.3239 2.5888 2.9027 3.1198 3.3232 3.5751 3.7550 4 1.6972 2.1561 2.4420 2.6965 2.9997 3.2103 3.4082 3.6541 3.8301 5 1.8106 2.2520 2.5286 2.7758 3.0715 3.2775 3.4716 3.7132 3.8864 6 1.8990 2.3272 2.5967 2.8384 3.1283 3.3309 3.5220 3.7603 3.9314 7 1.9711 2.3887 2.6527 2.8900 3.1753 3.3751 3.5638 3.7995 3.9690 8 2.0317 2.4407 2.7000 2.9337 3.2153 3.4127 3.5995 3.8331 4.0011 9 2.0838 2.4856 2.7411 2.9717 3.2500 3.4455 3.6307 3.8623 4.0292 10 2.1295 2.5250 2.7772 3.0052 3.2807 3.4745 3.6582 3.8883 4.0541 11 2.1701 2.5602 2.8094 3.0351 3.3082 3.5005 3.6830 3.9116 4.0765 12 2.2065 2.5918 2.8384 3.0621 3.3331 3.5241 3.7054 3.9328 4.0969 13 2.2395 2.6206 2.8649 3.0867 3.3558 3.5456 3.7259 3.9521 4.1155 14 2.2697 2.6469 2.8891 3.1093 3.3766 3.5653 3.7447 3.9699 4.1326 15 2.2975 2.6712 2.9115 3.1301 3.3959 3.5836 3.7622 3.9865 4.1485 16 2.3232 2.6937 2.9323 3.1495 3.4138 3.6007 3.7784 4.0019 4.1633 17 2.3471 2.7146 2.9516 3.1676 3.4306 3.6166 3.7936 4.0163 4.1772 18 2.3694 2.7342 2.9698 3.1846 3.4463 3.6315 3.8079 4.0298 4.1902 19 2.3904 2.7527 2.9868 3.2005 3.4611 3.6456 3.8214 4.0425 4.2025 20 2.4101 2.7700 3.0029 3.2156 3.4751 3.6589 3.8341 4.0546 4.2142 21 2.4287 2.7864 3.0181 3.2298 3.4883 3.6715 3.8461 4.0660 4.2252 22 2.4463 2.8020 3.0325 3.2434 3.5009 3.6835 3.8576 4.0769 4.2357 23 2.4631 2.8168 3.0463 3.2562 3.5128 3.6949 3.8685 4.0873 4.2457 24 2.4790 2.8309 3.0593 3.2685 3.5243 3.7058 3.8790 4.0972 4.2553 25 2.4941 2.8443 3.0718 3.2802 3.5352 3.7162 3.8889 4.1066 4.2644 26 2.5086 2.8572 3.0838 3.2915 3.5456 3.7262 3.8985 4.1157 4.2732 27 2.5225 2.8695 3.0953 3.3023 3.5557 3.7357 3.9077 4.1245 4.2816 28 2.5358 2.8813 3.1063 3.3126 3.5653 3.7449 3.9165 4.1328 4.2897 29 2.5486 2.8927 3.1168 3.3225 3.5746 3.7538 3.9250 4.1409 4.2975 30 2.5609 2.9036 3.1270 3.3321 3.5835 3.7623 3.9332 4.1487 4.3050 Table 2-2 One-sided tolerance factors for limits on sample minimum values

2.2.5 Modified Coefficient of Variation

A common problem with new material qualifications is that the initial specimens produced and tested do not contain all of the variability that will be encountered when the material is being produced in larger amounts over a lengthy period of time. This can result in setting basis values that are unrealistically high.

The modified Coefficient of Variation (CV) used in this report is in accordance with section 8.4.4 of CMH-17 Revision G. It is a method of adjusting the original basis values downward in anticipation of the expected additional variation. Composite materials are expected to have a CV of at least 6%. When the CV is less than 8%, a modification is made that adjusts the CV upwards.  .06  if CV  .04 CV Modified CV = CV *  .04if .04  CV .08 Equation 1 2  if CV  .08  CV

This is converted to percent by multiplying by 100%.

Page 12 of 49

139 April 11, 2016 NCP-RP-2016-001 NC

CV* is used to compute a modified standard deviation S*.

S** CV  X Equation 2

To compute the pooled standard deviation based on the modified CV:

k 2 n1 CV* X i  ii   * i1 S p  k Equation 3 ni 1  i1

The A-basis and B-basis values under the assumption of the modified CV method are computed by replacing S with S*.

When the basis values have been set using the modified CV method, we can use the modified CV to compute the equivalency test results.

Page 13 of 49

140 April 11, 2016 NCP-RP-2016-001 NC

3. Equivalency Test Results

The three NASA locations each had 23 different tests conducted. When combined, all tests had sufficient data for comparison purposes, but when examining individual locations only the modulus tests had sufficient data to meet the CMH-17 Rev G requirements, but the strength tests lacked sufficient data for the results to be considered conclusive. The Glenn location had six additional tests conducted for both the Low Vacuum panels and the Slow Ramp Panels. These results were evaluated separately and were not included in the combined dataset. All equivalency tests were performed with an α level of 5%.

Before the data from different locations can be combined for an equivalency test, it must be statistically tested to determine if the three locations produce similar enough product for the combined data to be considered unstructured, or having no significant differences between the test results from the different locations. When the data fail this test, an underlying assumption of the equivalency test is violated. The different locations will need to be evaluated individually for any property that shows differences between the locations.

The results of the equivalency comparisons are listed as ‘Pass’, ‘Fail’, or ‘Pass with Mod. CV’. ‘Pass with Mod CV’ refers to cases where the equivalency fails unless the modified coefficient of variation method is used. A minimum of eight samples from two separate panels and processing cycles is required for strength properties and a minimum of four specimens are required for modulus comparison tests. If the sample does not have an adequate number of specimens, this will be indicated with ‘Insufficient Data’ after the Pass or Fail indication. A summary of results for the combined locations is shown in Table 3-1, NASA Glenn is shown in Table 3-2, for NASA Langley in Table 3-3, and for NASA Marshall in Table 3-4.

Failures in table 3-1 through 3-5 are reported as "Failed by _._%". This percentage was computed by taking the ratio of the equivalency mean or minimum value to the modified CV limit for that value. Table 3-5 gives a rough scale for the relative severity of those failures.

Page 14 of 49

141 April 11, 2016 NCP-RP-2016-001 NC

Equivalency Test Results for NASA Glenn, Langley and Marshall Combined

Environmental Condition Normalized Test Property Data CTD RTD ETW

Strength Pass Pass Pass Longitudinal Yes Tension Pass with Failed by Modulus Pass Mod CV 0.5% Laminate Short Pass with No Strength Pass Beam Strength Mod CV Yes Strength Pass Pass Unnotched Compression * Data Not Yes Modulus Pass Combined * Data Not Strength Pass Pass Unnotched Combined Yes Tension * Data Not Failed by Modulus Pass Combined 0.2% Open Hole Failed by Failed by Yes Strength Compression 1.6% 5.8% Open Hole * Data Not Yes Strength Pass Pass Tension Combined * Statistically significant differences between locations indicate that data should not be combined. Table 3-1 Summary of Equivalency Test Results for NASA Combined Locations

Page 15 of 49

142 April 11, 2016 NCP-RP-2016-001 NC

Equivalency Test Results for Nasa Glenn M Cure Cycle with Hexcel 8552 IM7 M Cure Cycle

Environmental Condition Normalized Test Property Data CTD RTD ETD ETW Pass Pass Pass Strength Insufficient Insufficient Insufficient Longitudinal Yes Data Data Data Tension Pass with Modulus Pass Pass Mod CV Pass with Pass Laminate Short Mod CV No Strength Insufficient Beam Strength Insufficient Data Data Pass Pass Unnotched Yes Strength Insufficient Insufficient Compression Data Data (UNC1) Failed by Yes Modulus Pass 2.9% Pass Pass Pass Strength Insufficient Insufficient Insufficient Unnotched Tension Yes Data Data Data (UNT1) Modulus Pass Pass Pass

Failed by Failed by Open Hole 1.1% 5.1% Compression Yes Strength Insufficient Insufficient (OHC1) Data Data Pass Pass Pass Open Hole Tension Yes Strength Insufficient Insufficient Insufficient (OHT1) Data Data Data Table 3-2 Summary of Equivalency Test Results for NASA Glenn

Page 16 of 49

143 April 11, 2016 NCP-RP-2016-001 NC

Equivalency Test Results for NASA Langley M Cure Cycle with Hexcel 8552 IM7 M Cure Cycle

Environmental Condition Normalized Test Property Data CTD RTD ETD ETW Pass Pass Pass Strength Insufficient Insufficient Insufficient Longitudinal Yes Data Data Data Tension Pass with Pass with Modulus Pass Mod CV Mod CV Pass Pass Laminate Short No Strength Insufficient Insufficient Beam Strength Data Data Pass Pass Unnotched Yes Strength Insufficient Insufficient Compression Data Data (UNC1) Pass with Yes Modulus Pass Mod CV Pass Pass Pass Strength Insufficient Insufficient Insufficient Unnotched Tension Yes Data Data Data (UNT1) Modulus Pass Pass Pass

Pass with Failed by Open Hole Mod CV 4.8% Compression Yes Strength Insufficient Insufficient (OHC1) Data Data Pass Pass Pass Open Hole Tension Yes Strength Insufficient Insufficient Insufficient (OHT1) Data Data Data Table 3-3 Summary of Equivalency Test Results for NASA Langley

Page 17 of 49

144 April 11, 2016 NCP-RP-2016-001 NC

Equivalency Test Results for Nasa Marshall M Cure Cycle with Hexcel 8552 IM7 M Cure Cycle

Environmental Condition Normalized Test Property Data CTD RTD ETD ETW Pass Pass Pass Strength Insufficient Insufficient Insufficient Longitudinal Yes Data Data Data Tension Pass with Pass with Modulus Pass Mod CV Mod CV Failed by Pass Laminate Short 0.1% No Strength Insufficient Beam Strength Insufficient Data Data Pass Pass Unnotched Yes Strength Insufficient Insufficient Compression Data Data (UNC1) Pass with Yes Modulus Pass Mod CV Pass Pass Pass Strength Insufficient Insufficient Insufficient Unnotched Tension Yes Data Data Data (UNT1) Modulus Pass Pass Pass

Pass with Failed by Open Hole Mod CV 1.9% Compression Yes Strength Insufficient Insufficient (OHC1) Data Data Pass Pass Pass Open Hole Tension Yes Strength Insufficient Insufficient Insufficient (OHT1) Data Data Data Table 3-4 Summary of Equivalency Test Results for NASA Marshall

Description Modulus Strength Mild Failure % fail ≤ 4% % fail ≤ 5% Mild to Moderate Failure 4% < % fail ≤ 8% 5% < % fail ≤ 10% Moderate Failure 8% < % fail ≤ 12% 10%< % fail ≤ 15% Moderate to Severe Failure 12% < % fail ≤ 16% 15% < % fail ≤ 20% Severe Failure 16% < % fail ≤ 20% 20% < % fail ≤ 25% Extreme Failure 20% < % fail 25% < % fail Table 3-5 "% Failed" Results Scale

Page 18 of 49

145 April 11, 2016 NCP-RP-2016-001 NC

Graphical presentations of the combined location test results are shown in Figure 3-1 and Figure 3-2. In order to show different tests on the same graphical scale, all values are plotted as a percentage of the corresponding qualification mean. Figure 3-1 shows the strength means in the upper part of the chart using left axis and the strength minimums in the lower part of the chart using the right axis. This was done to avoid overlap of the two sets of data and equivalency criteria. Figure 3-2 shows the equivalency means plotted with the upper and lower equivalency criteria.

Strength Results as a Percentage of Qualification Mean Qual. Mean Equiv. Mean Lower Limits (Equiv. Mean) Mod CV Lower Limits (Equiv. Mean) Test Failure (Equiv. Mean) Qual. Mean Equiv. Min Lower Limits (Equiv. Min.) Mod CV Lower Limits (Equiv. Min.) Test Failure (Equiv. Min.)

110% 140%

100% 130%

90% 120%

80% 110%

70% 100% % of Mean 60% 90% % of Minimum

50% 80%

40% 70%

30% 60% RTD ETW RTD ETW CTD RTD ETW CTD RTD ETW CTD RTD ETW RTD ETW UNC1 OHC LT UNT1 OHT LSBS Normalized as measured

Figure 3-1 Summary of Combined Strength means and minimums compared to their respective Equivalence limits

Modulus, CPT and DMA Results as Percentage of Qualification Mean

Qual. Mean Equiv. Mean Upper Limits (Equiv. Mean) Lower Limit (Equiv. Mean) Mod CV Upper Limit (Equiv. Mean) Mod CV Lower Limit (Equiv. Mean)

115%

110%

105%

100% % of Mean

95%

90%

85% RTD ETW CTD RTD ETW CTD RTD ETW UNC1 LT UNT1

Figure 3-2 Summary of Combined Modulus means and Equivalence limits

Page 19 of 49

146 April 11, 2016 NCP-RP-2016-001 NC

4. Individual Test Results

4.1 Longitudinal (0°) Tension (LT)

The normalized LT data from all three locations could be combined. The combined dataset passed all equivalency tests for strength but failed equivalence for modulus in the ETW condition. All three locations failed due to the modulus values being too low. The individual locations passed with the use of the modified CV method, but the combined sample did not. This is due to the fact that larger samples have tighter acceptance limits.

Modified CV results could not be computed for the ETW strength data because that condition had a CV larger than 8%, which is not modified. However, the individual locations all passed the modulus equivalency tests with the use of the modified CV approach. There was insufficient strength data for results of the individual locations to be considered conclusive.

Statistics and analysis results are shown for the strength data CTD condition in Table 4-1, RTD condition in Table 4-2, ETW condition in Table 4-3 and for the modulus data CTD condition in Table 4-4, RTD Condition in Table 4-5 and ETW condition in Table 4-6.

Longitudinal Tension (LT) Qualification Glenn Langley Marshall Combined Strength CTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 357.389 363.658 368.170 358.027 363.285 Standard Deviation 12.620 12.879 11.948 6.836 10.756 Coefficient of Variation % 3.531 3.541 3.245 1.909 2.961 Minimum 325.692 349.131 354.456 348.530 348.530 Maximum 379.970 379.884 382.683 364.778 382.683 Number of Specimens 22 4 4 4 12 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 345.417 345.417 345.417 350.360 Minimum Acceptable Equiv. Sample Min 326.570 326.570 326.570 321.567 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV % 6.000 Minimum Acceptable Equiv. Sample Mean 337.048 337.048 337.048 345.447 Minimum Acceptable Equiv. Sample Min 305.024 305.024 305.024 296.524 Table 4-1 Longitudinal Tension Strength Results for CTD Condition

Page 20 of 49

147 April 11, 2016 NCP-RP-2016-001 NC

Longitudinal Tension (LT) Qualification Glenn Langley Marshall Combined Strength RTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 362.693 388.287 381.733 397.142 389.054 Standard Deviation 16.057 6.426 14.209 10.799 11.901 Coefficient of Variation % 4.427 1.655 3.722 2.719 3.059 Minimum 325.685 381.912 371.088 381.002 371.088 Maximum 392.322 397.057 402.585 403.821 403.821 Number of Specimens 18 4 4 4 12 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 347.462 347.462 347.462 353.751 Minimum Acceptable Equiv. Sample Min 323.483 323.483 323.483 317.118 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV % 6.214 Minimum Acceptable Equiv. Sample Mean 341.315 341.315 341.315 350.143 Minimum Acceptable Equiv. Sample Min 307.660 307.660 307.660 298.727 Table 4-2 Longitudinal Tension Strength Results for RTD Condition

Longitudinal Tension (LT) Strength Qualification Glenn Langley Marshall Combined ETW Data Data normalized with CPT Mean Strength (ksi) 333.504 356.367 358.081 366.531 360.005 Standard Deviation 38.823 11.526 13.232 12.901 12.462 Coefficient of Variation % 11.641 3.234 3.695 3.520 3.462 Minimum 244.533 340.091 338.169 348.106 338.169 Maximum 373.234 367.145 375.125 377.046 377.046 Number of Specimens 18 4 6 4 14 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 296.676 303.191 296.676 313.463 Minimum Acceptable Equiv. Sample Min 238.698 232.692 238.698 221.340 Table 4-3 Longitudinal Tension Strength Results for ETW Condition

Longitudinal Tension (LT) Qualification Glenn Langley Marshall Combined Modulus CTD Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 22.568 22.207 22.146 22.144 22.304 Standard Deviation 0.387 0.133 0.196 0.229 0.353 Coefficient of Variation % 1.717 0.601 0.885 1.036 1.581 Minimum 21.852 22.055 21.965 21.855 21.855 Maximum 23.219 22.343 22.418 22.391 23.116 Number of Specimens 22 4 4 4 12 RESULTS PASS FAIL FAIL PASS Passing Range for Modulus Mean 22.158 to 22.978 22.154 to 22.982 22.151 to 22.293 to 22.843 Student's t-statistic -1.818 -2.104 -2.100 -1.957 p-value of Student's t-statistic 0.082 0.046 0.046 0.059 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV% 6.000 Passing Range for Modulus Mean 21.146 to 23.990 21.145 to 23.991 21.144 to 21.752 to 23.384 Modified CV Student's t-statistic -0.524 -0.612 -0.615 -0.659 p-value of Student's t-statistic 0.605 0.546 0.545 0.515 Table 4-4 Longitudinal Tension Modulus Results for CTD Condition

Page 21 of 49

148 April 11, 2016 NCP-RP-2016-001 NC

Longitudinal Tension (LT) Qualification Glenn Langley Marshall Combined Modulus RTD Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 22.987 22.561 22.350 22.565 22.354 Standard Deviation 0.812 0.501 0.543 0.069 0.352 Coefficient of Variation % 3.532 2.219 2.429 0.308 1.576 Minimum 20.707 22.117 21.875 22.507 21.875 Maximum 23.941 23.116 23.074 22.660 23.074 Number of Specimens 18 4 4 4 12 RESULTS PASS PASS PASS FAIL Passing Range for Modulus Mean 22.095 to 23.878 22.090 to 23.883 22.123 to 23.85 22.475 to 23.498 Student's t-statistic -0.996 -1.481 -1.018 -2.535 p-value of Student's t-statistic 0.331 0.154 0.321 0.017 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV% 6.000 Passing Range for Modulus Mean 21.504 to 24.470 21.501 to 24.473 21.520 to 22.149 to 23.824 Modified CV Student's t-statistic -0.599 -0.894 -0.600 -1.548 p-value of Student's t-statistic 0.556 0.382 0.555 0.133 Table 4-5 Longitudinal Tension Modulus Results for RTD Condition

Longitudinal Tension (LT) Qualification Glenn Langley Marshall Combined Modulus ETW Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 24.001 23.249 23.160 22.847 23.096 Standard Deviation 0.557 0.393 0.367 0.203 0.354 Coefficient of Variation % 2.321 1.690 1.583 0.887 1.531 Minimum 23.222 22.787 22.693 22.652 22.652 Maximum 25.578 23.744 23.663 23.087 23.744 Number of Specimens 29 4 6 4 14 RESULTS FAIL FAIL FAIL FAIL Passing Range for Modulus Mean 23.410 to 24.592 23.515 to 24.487 23.421 to 23.671 to 24.33 Student's t-statistic -2.596 -3.521 -4.059 -5.545 p-value of Student's t-statistic 0.014 0.001 0.0003 0.000002 PASS with PASS with PASS with FAIL MOD CV RESULTS MOD CV MOD CV MOD CV Modified CV% 6.000 Passing Range for Modulus Mean 22.506 to 25.495 22.783 to 25.218 22.510 to 23.208 to 24.794 Modified CV Student's t-statistic -1.026 -1.405 -1.579 -2.305 p-value of Student's t-statistic 0.313 0.169 0.124 0.026 Table 4-6 Longitudinal Tension Modulus Results for ETW Condition

The LT modulus data for the combined data in the RTD environment failed the equivalency test because the sample mean value (22.354) is below the lower acceptance limit (22.475). The equivalency sample mean value is 99.46% of the lower limit of acceptable values. Under the assumption of the modified CV method, the modulus data from the RTD environment passed the equivalence test.

The LT modulus data for the combined data in the ETW environment failed the equivalency test because the sample mean value (23.096) is below the lower acceptance limit (23.671). The equivalency sample mean value is 97.57% of the lower limit of acceptable values. Under the assumption of the modified CV method, it was 99.52% of the lower limit of acceptable values (23.208).

Page 22 of 49

149 April 11, 2016 NCP-RP-2016-001 NC

The LT modulus data for Glenn in ETW environment failed the equivalency test because the sample mean value (23.249) is below the lower acceptance limit (23.410). The equivalency sample mean value is 99.31% of the lower limit of acceptable values. Under the assumption of the modified CV method, the modulus data from the ETW environment passed the equivalence test.

The LT modulus data for Langley in the CTD environment failed the equivalency test because the sample mean value (22.146) is below the lower acceptance limit (22.154). The equivalency sample mean value is 99.96% of the lower limit of acceptable values. Under the assumption of the modified CV method, the modulus data from the CTD environment passed the equivalence test.

The LT modulus data for Langley in the ETW environment failed the equivalency test because the sample mean value (23.160) is below the lower acceptance limit (23.515). The equivalency sample mean value is 98.49% of the lower limit of acceptable values. Under the assumption of the modified CV method, the modulus data from the ETW environment passed the equivalence test.

The LT modulus data for Marshall in the CTD environment failed the equivalency test because the sample mean value (22.144) is below the lower acceptance limit (22.151). The equivalency sample mean value is 99.97% of the lower limit of acceptable values. Under the assumption of the modified CV method, the modulus data from the CTD environment passed the equivalence test.

The LT modulus data for Marshall in the ETW environment failed the equivalency test because the sample mean value (22.847) is below the lower acceptance limit (23.421). The equivalency sample mean value is 97.55% of the lower limit of acceptable values. Under the assumption of the modified CV method, the modulus data from the ETW environment passed the equivalence test.

The LT strength and modulus values and equivalency acceptance limits for the CTD condition in Figure 4-1, for the RTD condition in Figure 4-2, and for the ETW condition in Figure 4-3. The acceptance limits shown are for the individual location samples, not the combined sample which will have slightly tighter acceptance limits due to the larger sample size.

The graphs indicate that the NASA samples (all conditions) have slightly lower modulus values than the qualification sample did, but there are no obvious differences between the three locations.

Page 23 of 49

150 April 11, 2016 NCP-RP-2016-001 NC

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Longitudinal Tension CTD Condition 390

380

370

360

350

340 Strength (ksi) 330

320

310

300 21 21.5 22 22.5 23 23.5 24 24.5 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Mod CV Min. Ind. Spec. Str. Acceptance Limit Mod CV Min. Ind. Spec. Str. Acceptance Limit Figure 4-1 Longitudinal Tension Values, Averages and Acceptance limits for CTD Condition

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Longitudinal Tension RTD Condition 420

400

380

360 Strength (ksi) 340

320

300 21 21.5 22 22.5 23 23.5 24 24.5 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Min. Ind. Spec. Str. Acceptance Limit Mod CV Min. Ind. Spec. Str. Acceptance Limit Figure 4-2 Longitudinal Tension Values, Averages and Acceptance limits for RTD Condition

Page 24 of 49

151 April 11, 2016 NCP-RP-2016-001 NC

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Longitudinal Tension ETW Condition 390

370

350

330

310

Strength (ksi) 290

270

250

230 22 22.5 23 23.5 24 24.5 25 25.5 26 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Min. Ind. Spec. Str. Acceptance Limit Figure 4-3 Longitudinal Tension Values, Averages and Acceptance limits for ETW Condition

Page 25 of 49

152 April 11, 2016 NCP-RP-2016-001 NC

4.2 Quasi Isotropic (“25/50/25”) Unnotched Tension (UNT1)

The UNT1 normalized strength data from all three locations could be combined for CTD and RTD but not the ETW condition. The UNT1 normalized modulus data from all three locations could be combined for the RTD and ETW but not the CTD condition.

The only equivalency test failure was for the combined modulus dataset in the RTD condition. This is surprising because each of the individual locations passed equivalency for this property and condition. However, they are all on the low side and the tighter acceptance criteria of the larger sample resulted in an equivalency failure for the combined sample.

Statistics and analysis results are shown for the strength data CTD condition in Table 4-7, RTD condition in Table 4-8, ETW condition in Table 4-9 and for the modulus data CTD condition in Table 4-10, RTD Condition in Table 4-11and ETW condition in Table 4-12.

The data and acceptance limits are shown graphically for the CTD condition in Figure 4-4, the RTD condition in Figure 4-5, and the ETW condition in Figure 4-6.

Unnotched Tension (UNT1) Qualification Glenn Langley Marshall Combined Strength CTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 99.348 109.862 107.068 109.487 108.806 Standard Deviation 3.442 3.298 1.502 1.747 2.526 Coefficient of Variation % 3.464 3.002 1.403 1.595 2.322 Minimum 91.601 104.504 105.336 107.820 104.504 Maximum 105.840 113.498 108.892 112.351 113.498 Number of Specimens 16 6 6 6 18 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 96.660 96.660 96.660 97.778 Minimum Acceptable Equiv. Sample Min 90.411 90.411 90.411 89.127 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV % 6.000 Minimum Acceptable Equiv. Sample Mean 94.693 94.693 94.693 97.778 Minimum Acceptable Equiv. Sample Min 83.869 83.869 83.869 89.127 Table 4-7 UNT1 Strength Results for CTD Condition

Page 26 of 49

153 April 11, 2016 NCP-RP-2016-001 NC

Unnotched Tension (UNT1) Qualification Glenn Langley Marshall Combined Strength RTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 104.685 105.314 108.002 107.372 106.896 Standard Deviation 7.276 7.065 1.534 1.647 4.191 Coefficient of Variation % 6.950 6.708 1.420 1.534 3.921 Minimum 89.563 92.133 105.873 105.416 92.133 Maximum 113.712 109.828 109.469 109.930 109.930 Number of Specimens 16 6 6 6 18 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 99.004 99.004 99.004 101.367 Minimum Acceptable Equiv. Sample Min 85.792 85.792 85.792 83.077 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV % 7.638 Minimum Acceptable Equiv. Sample Mean 98.575 98.575 98.575 101.367 Minimum Acceptable Equiv. Sample Min 84.365 84.365 84.365 83.077 Table 4-8 UNT1 Strength Results for RTD Condition

Unnotched Tension (UNT1) Qualification Glenn Langley Marshall Combined Strength ETW Data Data normalized with CPT 0.0072 Mean Strength (ksi) 112.461 108.359 116.139 110.903 111.800 Standard Deviation 5.606 3.250 2.748 2.785 4.326 Coefficient of Variation % 4.985 2.999 2.366 2.511 3.869 Minimum 101.642 103.102 112.493 107.472 103.102 Maximum 119.290 111.969 120.027 114.983 120.027 Number of Specimens 16 6 6 6 18 RESULTS PASS PASS PASS Statistical Minimum Acceptable Equiv. Sample Mean 108.084 108.084 108.084 Differences Minimum Acceptable Equiv. Sample Min 97.904 97.904 97.904 between PASS with PASS with PASS with MOD CV RESULTS Locations MOD CV MOD CV MOD CV preclude Modified CV % 6.803 Minimum Acceptable Equiv. Sample Mean 106.760 106.760 106.760 combining Minimum Acceptable Equiv. Sample Min 93.501 93.501 93.501 results Table 4-9 UNT1 Strength Results for ETW Condition

Page 27 of 49

154 April 11, 2016 NCP-RP-2016-001 NC

Unnotched Tension (UNT1) Qualification Glenn Langley Marshall Combined Modulus CTD Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 8.353 8.178 8.414 8.343 8.312 Standard Deviation 0.309 0.114 0.110 0.038 0.135 Coefficient of Variation % 3.696 1.398 1.307 0.455 1.622 Minimum 7.295 8.013 8.297 8.310 8.013 Maximum 8.745 8.362 8.619 8.414 8.619 Number of Specimens 16 6 6 6 18 RESULTS PASS PASS PASS Passing Range for Modulus Mean 8.080 to 8.626 8.081 to 8.626 8.086 to 8.621 Statistical Student's t-statistic -1.341 0.464 -0.084 Differences p-value of Student's t-statistic 1.95E-01 0.6478 0.93418 between PASS with PASS with PASS with Locations MOD CV RESULTS MOD CV MOD CV MOD CV preclude Modified CV% 6.000 Passing Range for Modulus Mean 7.916 to 8.791 7.917 to 8.790 0.000 combining Modified CV Student's t-statistic -0.838 0.289 -0.052 results p-value of Student's t-statistic 0.412 0.775 0.959 Table 4-10 UNT1 Modulus Results for CTD Condition

Unnotched Tension (UNT1) Qualification Glenn Langley Marshall Combined Modulus RTD Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 8.390 7.923 8.186 8.070 8.060 Standard Deviation 0.480 0.455 0.165 0.101 0.290 Coefficient of Variation % 5.727 5.743 2.021 1.249 3.601 Minimum 7.277 7.049 8.013 7.964 7.049 Maximum 8.984 8.394 8.494 8.223 8.494 Number of Specimens 16 6 6 6 18 RESULTS PASS PASS PASS FAIL Passing Range for Modulus Mean 7.917 to 8.864 7.967 to 8.814 7.972 to 8.809 8.116 to 8.664 Student's t-statistic -2.060 -1.004 -1.598 -2.461 p-value of Student's t-statistic 5.27E-02 0.3275 0.12583 0.0195 PASS with PASS with PASS with FAIL MOD CV RESULTS MOD CV MOD CV MOD CV Modified CV% 6.863 Passing Range for Modulus Mean 7.843 to 8.938 7.885 to 8.895 7.890 to 8.891 8.077 to 8.703 Modified CV Student's t-statistic -1.782 -0.842 -1.336 -2.151 p-value of Student's t-statistic 0.090 0.410 0.197 0.039 Table 4-11 UNT1 Modulus Results for RTD Condition

Page 28 of 49

155 April 11, 2016 NCP-RP-2016-001 NC

Unnotched Tension (UNT1) Qualification Glenn Langley Marshall Combined Modulus ETW Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 7.988 7.976 8.024 7.923 7.974 Standard Deviation 0.412 0.083 0.145 0.130 0.122 Coefficient of Variation % 5.162 1.038 1.811 1.641 1.536 Minimum 7.069 7.836 7.908 7.809 7.809 Maximum 8.514 8.049 8.215 8.155 8.215 Number of Specimens 17 6 6 6 18 RESULTS PASS PASS PASS PASS Passing Range for Modulus Mean 7.630 to 8.346 7.626 to 8.35 7.627 to 8.349 7.782 to 8.195 Student's t-statistic -0.072 0.208 -0.374 -0.135 p-value of Student's t-statistic 9.43E-01 0.8374 0.71188 0.89331 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV% 6.581 Passing Range for Modulus Mean 7.533 to 8.443 7.530 to 8.447 7.531 to 8.446 7.729 to 8.247 Modified CV Student's t-statistic -0.057 0.164 -0.295 -0.108 p-value of Student's t-statistic 0.955 0.871 0.771 0.915 Table 4-12 UNT1 Modulus Results for ETW Condition

The UNT1 RTD modulus data in ETW environment failed the equivalency test because the sample mean value (8.060) is below the lower acceptance limit (8.116). The equivalency sample mean value is 99.30% of the lower limit of acceptable values. Under the assumption of the modified CV method, it was 99.78% of the lower limit of acceptable values (8.077).

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Quasi Isotropic Unnotched Tension CTD Condition 125

120

115

110

105

100 Strength (ksi) 95

80 7 7.2 7.4 7.6 7.8 8 8.2 8.4 8.6 8.8 9 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Min. Ind. Spec. Str. Acceptance Limit Mod CV Min. Ind. Spec. Str. Acceptance Limit Outlier Figure 4-4 UNT1 Values, Averages and Acceptance limits for CTD Condition

Page 29 of 49

156 April 11, 2016 NCP-RP-2016-001 NC

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Quasi Isotropic Unnotched Tension RTD Condition 125

120

115

110

105

100 Strength (ksi) 95

80 7 7.2 7.4 7.6 7.8 8 8.2 8.4 8.6 8.8 9 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Min. Ind. Spec. Str. Acceptance Limit Mod CV Min. Ind. Spec. Str. Acceptance Limit Outlier Figure 4-5 UNT1 Values, Averages and Acceptance limits for RTD Condition

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Quasi Isotropic Unnotched Tension ETW Condition 125

120

115

110

105

100 Strength (ksi) 95

80 7 7.2 7.4 7.6 7.8 8 8.2 8.4 8.6 8.8 9 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Min. Ind. Spec. Str. Acceptance Limit Mod CV Min. Ind. Spec. Str. Acceptance Limit Figure 4-6 UNT1 Values, Averages and Acceptance limits for ETW Condition

Page 30 of 49

157 April 11, 2016 NCP-RP-2016-001 NC

4.3 Quasi Isotropic “25/50/25” Open-Hole Tension 1 (OHT1)

The OHT1 normalized strength data from all three locations could be combined for CTD and ETW but not the RTD condition. There were no equivalency test failures for the OHT1 data, either for the individual locations or the combined dataset.

Statistics and analysis results are shown for the strength data CTD condition in Table 4-13, RTD condition in Table 4-14, ETW condition in Table 4-15. The data and acceptance limits are shown graphically for all condition in Figure 4-7.

Open Hole Tension (OHT1) Qualification Glenn Langley Marshall Combined Strength CTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 57.754 64.584 63.564 62.631 63.593 Standard Deviation 2.433 1.471 1.756 1.352 1.660 Coefficient of Variation % 4.213 2.277 2.763 2.159 2.610 Minimum 53.645 63.046 61.618 60.242 60.242 Maximum 62.524 66.763 66.570 64.241 66.763 Number of Specimens 19 6 6 6 18 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 55.855 55.855 55.855 56.645 Minimum Acceptable Equiv. Sample Min 51.436 51.436 51.436 50.528 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV % 6.106 Minimum Acceptable Equiv. Sample Mean 55.001 55.001 55.001 56.146 Minimum Acceptable Equiv. Sample Min 48.597 48.597 48.597 47.281 Table 4-13 OHT1 Strength Results for CTD

Open Hole Tension (OHT1) Qualification Glenn Langley Marshall Combined Strength RTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 59.003 65.988 62.410 64.072 64.156 Standard Deviation 2.350 1.846 1.117 1.697 2.117 Coefficient of Variation % 3.982 2.798 1.789 2.649 3.299 Minimum 54.120 62.985 61.142 62.276 61.142 Maximum 64.610 68.375 63.946 66.602 68.375 Number of Specimens 19 6 6 6 18 RESULTS PASS PASS PASS Statistical Minimum Acceptable Equiv. Sample Mean 57.169 57.169 57.169 Differences Minimum Acceptable Equiv. Sample Min 52.902 52.902 52.902 between PASS with PASS with PASS with MOD CV RESULTS Locations MOD CV MOD CV MOD CV preclude Modified CV % 6.000 Minimum Acceptable Equiv. Sample Mean 56.239 56.239 56.239 combining Minimum Acceptable Equiv. Sample Min 49.811 49.811 49.811 results Table 4-14 OHT1 Strength Results for RTD

Page 31 of 49

158 April 11, 2016 NCP-RP-2016-001 NC

Open Hole Tension (OHT1) Qualification Glenn Langley Marshall Combined Strength ETW Data Data normalized with CPT 0.0072 Mean Strength (ksi) 66.966 69.241 68.990 69.356 69.196 Standard Deviation 2.850 4.799 1.680 2.368 3.046 Coefficient of Variation % 4.255 6.931 2.436 3.414 4.402 Minimum 62.154 63.279 66.791 67.399 63.279 Maximum 72.587 76.500 70.669 73.603 76.500 Number of Specimens 20 6 6 6 18 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 64.742 64.742 64.742 65.667 Minimum Acceptable Equiv. Sample Min 59.567 59.567 59.567 58.504 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV % 6.128 Minimum Acceptable Equiv. Sample Mean 63.763 63.763 66.791 65.095 Minimum Acceptable Equiv. Sample Min 56.311 56.311 56.311 54.780 Table 4-15 OHT1 Strength Results for ETW

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Quasi Isotropic Open Hole Tenion(OHT1) 100

50 ksi

0 CTD RTD ETW Environment Glenn Langley Marshall Glenn Averages Langley Averages Marshall Averages CTD Acceptance Limit RTD Acceptance Limit CTD Mod CV Acceptance Limit RTD Mod CV Acceptance Limit ETW Acceptance Limit ETW Mod CV Acceptance Limit

Figure 4-7 OHT1 Values, Averages and Acceptance limits

Page 32 of 49

159 April 11, 2016 NCP-RP-2016-001 NC

4.4 Laminate Short-Beam Strength (LSBS)

The Laminate Short-Beam Strength data is not normalized. The LSBS data from all three locations could be combined. The combined data passed all equivalency tests, although the ETW data required the use of the modified CV approach. The individual locations all passed the equivalency tests for the RTD condition, but Glenn and Marshall failed the ETW equivalency tests. Glenn passed for the ETW condition with the use of the modified CV approach but Marshall did not. However, there was insufficient strength data for results of the individual locations to be considered conclusive.

Statistics and analysis results are shown for the RTD condition in Table 4-5 and for the ETW condition in Table 4-6.

Laminate Short Beam Qualification Glenn Langley Marshall Combined Strength (LSBS) RTD Data Data as measured (not normalized) Mean Strength (ksi) 12.129 12.446 12.033 12.399 12.301 Standard Deviation 0.831 0.398 0.466 0.474 0.459 Coefficient of Variation % 6.851 3.199 3.871 3.822 3.732 Minimum 9.550 11.987 11.423 11.687 11.423 Maximum 12.983 13.109 12.548 12.892 13.109 Number of Specimens 21 7 6 6 19 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 11.527 11.480 11.480 11.760 Minimum Acceptable Equiv. Sample Min 9.925 9.971 9.971 9.647 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV % 7.425 Minimum Acceptable Equiv. Sample Mean 11.476 11.426 11.426 11.729 Minimum Acceptable Equiv. Sample Min 9.740 9.790 9.790 9.439 Table 4-16 Laminate Short-Beam Strength Results for RTD

Laminate Short Beam Qualification Glenn Langley Marshall Combined Strength (LSBS) ETW Data Data as measured (not normalized) Mean Strength (ksi) 6.991 6.840 6.959 6.658 6.819 Standard Deviation 0.255 0.276 0.162 0.210 0.244 Coefficient of Variation % 3.646 4.040 2.325 3.149 3.573 Minimum 6.635 6.296 6.696 6.277 6.277 Maximum 7.699 7.050 7.150 6.850 7.150 Number of Specimens 19 6 6 6 18 RESULTS FAIL PASS FAIL FAIL Minimum Acceptable Equiv. Sample Mean 6.792 6.792 6.792 6.874 Minimum Acceptable Equiv. Sample Min 6.329 6.329 6.329 6.234 PASS with PASS with PASS with MOD CV RESULTS FAIL MOD CV MOD CV MOD CV Modified CV % 6.000 Minimum Acceptable Equiv. Sample Mean 6.663 6.663 6.663 6.799 Minimum Acceptable Equiv. Sample Min 5.901 5.901 5.901 5.745 Table 4-17 Laminate Short-Beam Strength Results for ETW

The LSBS data for the combined data in the ETW environment failed equivalence due sample minimum being too low. The equivalency minimum value is (6.819) is 99.19% of

Page 33 of 49

160 April 11, 2016 NCP-RP-2016-001 NC the lowest acceptable minimum value (6.874). Under the assumption of the modified CV method, the ETW environment passed the equivalence test.

The LSBS data for Glenn in the ETW environment failed equivalence due sample minimum being too low. The equivalency minimum value is (6.296) is 99.48% of the lowest acceptable minimum value (6.329). Under the assumption of the modified CV method, the ETW environment passed the equivalence test.

The LSBS data for Marshall in the ETW environment failed equivalence due to both the sample mean and sample minimum being too low. The equivalency sample mean (6.658) is 98.03% of the minimum acceptable mean value (6.792). The equivalency sample minimum (6.277) is 99.19% of the lowest acceptable minimum value (6.329). Under the assumption of the modified CV method, the equivalency sample mean is 99.92% of the lowest acceptable mean value (6.663). The equivalency sample minimum passed the test.

The LSBS strength means, individual specimen results and acceptance limits are shown for all labs in Figure 4-8.

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Laminate Short Beam Strength (LSBS)

8 ksi

0 RTD ETW Environment Glenn Langley Marshall Glenn Averages Langley Averages Marshall Averages RTD Acceptance Limit ETW Acceptance Limit RTD Mod CV Acceptance Limit ETW Mod CV Acceptance Limit Figure 4-8 Laminate Short-Beam Strength means, minimums and Acceptance limits for NASA Labs

Page 34 of 49

161 April 11, 2016 NCP-RP-2016-001 NC

4.5 Quasi Isotropic (“25/50/25”) Unnotched Compression 1 (UNC1)

The UNC1 data from all three locations could be combined for strength properties and modulus for the RTD condition, but there were statistically significant differences between the locations for modulus values in the ETW condition. The combined dataset passed the equivalency tests for strength and for the modulus RTD condition.

The three locations all failed equivalency tests for modulus in the ETW condition due to the modulus values being too high. Although the Langley and Marshall locations were able to pass equivalency with the use of the modified CV approach, Glenn was not. In fact, all of the modulus values from Glenn for the ETW condition fell above the upper acceptance limit.

Modified CV results could not be computed for the strength data because both conditions had a CV larger than 8%, which is not modified.

Statistics and analysis results are shown for strength RTD condition in Table 4-18, and for strength in the ETW condition in Table 4-19, for the modulus RTD condition in Table 4-20, and for modulus in the ETW condition in Table 4-21.

Unnotched Compression Qualification Glenn Langley Marshall Combined (UNC1) Strength RTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 87.045 94.483 92.219 94.951 92.389 Standard Deviation 8.111 7.341 2.075 3.160 4.733 Coefficient of Variation % 9.318 7.769 2.250 3.328 5.123 Minimum 68.065 84.038 88.961 91.952 80.740 Maximum 97.037 100.706 94.741 98.458 100.706 Number of Specimens 16 4 6 6 24 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 79.351 80.712 80.712 83.836 Minimum Acceptable Equiv. Sample Min 67.239 65.984 65.984 62.232

Table 4-18 UNC1 Strength Results for RTD

Unnotched Compression Qualification Glenn Langley Marshall Combined (UNC1) Strength ETW Data Data normalized with CPT 0.0072 Mean Strength (ksi) 57.675 60.488 56.413 60.336 57.701 Standard Deviation 6.355 7.683 4.257 1.674 4.967 Coefficient of Variation % 11.019 12.702 7.546 2.774 8.608 Minimum 48.716 54.063 51.380 58.131 48.189 Maximum 72.226 69.852 63.319 62.169 69.852 Number of Specimens 30 4 6 5 24 RESULTS PASS PASS PASS PASS Minimum Acceptable Equiv. Sample Mean 51.647 52.713 52.257 55.161 Minimum Acceptable Equiv. Sample Min 42.155 41.172 41.605 38.232

Table 4-19 UNC1 Strength Results for ETW

Page 35 of 49

162 April 11, 2016 NCP-RP-2016-001 NC

Unnotched Compression Qualification Glenn Langley Marshall Combined (UNC1) Modulus RTD Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 7.857 7.955 7.597 7.662 7.711 Standard Deviation 0.373 0.360 0.099 0.035 0.227 Coefficient of Variation % 4.749 4.526 1.307 0.451 1.801 Minimum 6.890 7.445 7.474 7.636 6.852 Maximum 8.407 8.274 7.706 7.714 7.344 Number of Specimens 16 4 6 6 16 RESULTS PASS PASS PASS PASS Passing Range for Modulus Mean 7.421 to 8.293 7.531 to 8.183 7.534 to 8.18 7.634 to 8.08 Student's t-statistic 0.473 -1.661 -1.256 -1.337 p-value of Student's t-statistic 0.642 0.112 0.223 0.191 PASS with PASS with PASS with PASS with MOD CV RESULTS MOD CV MOD CV MOD CV MOD CV Modified CV% 6.374 Passing Range for Modulus Mean 7.293 to 8.421 7.421 to 8.293 7.424 to 8.290 7.576 to 8.138 Modified CV Student's t-statistic 0.365 -1.244 -0.937 -1.062 p-value of Student's t-statistic 0.719 0.228 0.360 0.297

Table 4-20 UNC1 Modulus Results for RTD

Unnotched Compression Qualification Glenn Langley Marshall Combined (UNC1) Modulus ETW Data Data normalized with CPT 0.0072 Mean Modulus (Msi) 7.126 7.806 7.375 7.464 7.519 Standard Deviation 0.128 0.123 0.058 0.090 0.201 Coefficient of Variation % 1.801 1.582 0.783 1.206 2.666 Minimum 6.852 7.621 7.296 7.353 7.296 Maximum 7.344 7.878 7.443 7.578 7.878 Number of Specimens 16 4 6 5 15 RESULTS FAIL FAIL FAIL Passing Range for Modulus Mean 6.977 to 7.276 7.012 to 7.241 6.996 to 7.256 Statistical Student's t-statistic 9.528 4.528 5.429 Differences p-value of Student's t-statistic 1.87E-08 0.0002 0.00003 between PASS with PASS with FAIL Locations MOD CV RESULTS MOD CV MOD CV preclude Modified CV% 6.000 Passing Range for Modulus Mean 6.664 to 7.589 6.755 to 7.497 0.000 combining Modified CV Student's t-statistic 3.088 1.400 1.723 results p-value of Student's t-statistic 0.006 0.177 0.101

Table 4-21 UNC1 Modulus Results for ETW

The UNC1 modulus data for the Glenn ETW environment failed equivalence due to the sample mean being too high. The equivalency sample mean (7.806) is 107.28% of the maximum acceptable mean value (7.276). Under the assumption of the modified CV method, the equivalency sample mean is 102.85% of the highest acceptable mean value (7.589).

The UNC1 modulus data for the Langley ETW environment failed equivalence due to sample mean being too high. The equivalency sample mean (7.375) is 101.85% of the

Page 36 of 49

163 April 11, 2016 NCP-RP-2016-001 NC

maximum acceptable mean value (7.241). Under the assumption of the modified CV method, the equivalency sample mean passed the test.

The UNC1 modulus data for the Marshall ETW environment failed equivalence due to sample mean being too high. The equivalency sample mean (7.464) is 102.86% of the maximum acceptable mean value (7.256). Under the assumption of the modified CV method, the equivalency sample mean passed the test.

The UNC1 strength and modulus values and equivalency acceptance limits for the RTD condition in Figure 4-9, and for the ETW condition in Figure 4-10. The acceptance limits shown are for the individual location samples, not the combined sample which will have slightly tighter acceptance limits due to the larger sample size.

The graphs indicate that the Glenn Lab samples have higher modulus values than the other two locations, although the differences are not statistically significant for the RTD condition.

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Quasi Isotropic Unnotched Compression RTD Condition 110

100

80 Strength (ksi) 70

50 7 7.2 7.4 7.6 7.8 8 8.2 8.4 8.6 8.8 9 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Min. Ind. Spec. Str. Acceptance Limit Figure 4-9 UNC1 means, minimums and Acceptance limits for RTD Condition

Page 37 of 49

164 April 11, 2016 NCP-RP-2016-001 NC

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Quasi Isotropic Unnotched Compression ETW Condition 80

50 Strength (ksi)

30 6.5 6.7 6.9 7.1 7.3 7.5 7.7 7.9 8.1 8.3 8.5 Modulus (Msi)

Glenn Langley Marshall Glenn Average Langley Average Marshall Average Acceptance Limits for Averages Mod CV Acceptance Limits for Averages Min. Ind. Spec. Str. Acceptance Limit Figure 4-10 UNC1 means, minimums and Acceptance limits for ETW Condition

Page 38 of 49

165 April 11, 2016 NCP-RP-2016-001 NC

4.6 Quasi Isotropic “25/50/25” Open-Hole Compression 1 (OHC1)

The OHC1 normalized strength data from all three locations could be combined for both the RTD and ETW conditions. The combined datasets for OHC1 normalized strength data failed equivalency tests for the both RTD and ETW conditions. This was not surprising given that all three locations failed equivalency for both conditions, although the Langley and Marshall locations passed for the RTD condition with the use of the modified CV method.

Statistics and analysis results are shown for strength RTD condition in Table 4-22, and for strength in the ETW condition in Table 4-23.

The OHC1 strength means, individual specimen results and acceptance limits are shown for all labs in Figure 4-11.

Open Hole Compression Qualification Glenn Langley Marshall Combined (OHC1) Strength RTD Data Data normalized with CPT 0.0072 Mean Strength (ksi) 49.083 45.792 47.082 47.503 46.918 Standard Deviation 1.793 0.973 1.379 1.193 1.335 Coefficient of Variation % 3.653 2.124 2.928 2.512 2.846 Minimum 43.909 44.449 46.004 45.898 44.449 Maximum 50.993 46.741 49.699 49.343 49.699 Number of Specimens 19 4 6 6 16 RESULTS FAIL FAIL FAIL FAIL Minimum Acceptable Equiv. Sample Mean 47.383 47.683 47.683 48.217 Minimum Acceptable Equiv. Sample Min 44.705 44.428 44.428 43.826 PASS with PASS with MOD CV RESULTS FAIL FAIL MOD CV MOD CV Modified CV % 6.000 6.000 6.000 6.000 Minimum Acceptable Equiv. Sample Mean 46.290 46.784 46.784 47.660 Minimum Acceptable Equiv. Sample Min 41.892 41.436 41.436 40.448 Table 4-22 OHC1 Strength Results For RTD Condition

Open Hole Compression Qualification Glenn Langley Marshall Combined (OHC1) Strength ETW Data Data normalized with CPT 0.0072 Mean Strength (ksi) 35.515 31.767 32.214 33.186 32.467 Standard Deviation 1.445 0.444 0.766 0.910 0.935 Coefficient of Variation % 4.069 1.397 2.377 2.743 2.879 Minimum 33.080 31.130 31.638 31.785 31.130 Maximum 38.956 32.157 33.633 34.367 34.367 Number of Specimens 19 4 6 6 16 RESULTS FAIL FAIL FAIL FAIL Minimum Acceptable Equiv. Sample Mean 34.144 34.387 34.387 34.817 Minimum Acceptable Equiv. Sample Min 31.986 31.763 31.763 31.278 MOD CV RESULTS FAIL FAIL FAIL FAIL Modified CV % 6.034 Minimum Acceptable Equiv. Sample Mean 33.482 33.842 33.842 34.479 Minimum Acceptable Equiv. Sample Min 30.282 29.950 29.950 29.231 Table 4-23 OHC1 Strength Results For ETW Condition

Page 39 of 49

166 April 11, 2016 NCP-RP-2016-001 NC

The OHC1 combined strength data for the RTD environment failed equivalence due to sample mean being too low. The equivalency sample mean (46.918) is 97.31% of the minimum acceptable mean value (48.217). The equivalency sample minimum passed the test. Under the assumption of the modified CV method, the equivalency sample mean is 98.44% of the lowest acceptable mean value (47.660).

The OHC1 combined strength data for the ETW environment failed equivalence due to both the sample mean and sample minimum being too low. The equivalency sample mean (32.467) is 93.25% of the minimum acceptable mean value (34.817). The equivalency sample minimum (31.130) is 99.53% of the lowest acceptable minimum value (31.278). Under the assumption of the modified CV method, the equivalency sample mean is 94.16% of the lowest acceptable mean value (34.479). The equivalency sample minimum passed the test.

The Glenn OHC1 strength data for the RTD environment failed equivalence due to both the sample mean and sample minimum being too low. The equivalency sample mean (45.792) is 96.64% of the minimum acceptable mean value (47.383). The equivalency sample minimum (44.449) is 99.43% of the lowest acceptable minimum value (44.705). Under the assumption of the modified CV method, the equivalency sample mean is 98.93% of the lowest acceptable mean value (46.29). The equivalency sample minimum passed the test.

The Glenn OHC1 strength data for the ETW environment failed equivalence due to both the sample mean and sample minimum being too low. The equivalency sample mean (31.767) is 93.04% of the minimum acceptable mean value (34.144). The equivalency sample minimum (31.130) is 97.32% of the lowest acceptable minimum value (31.986). Under the assumption of the modified CV method, the equivalency sample mean is 94.88% of the lowest acceptable mean value (33.482). The equivalency sample minimum passed the test.

The Langley OHC1 strength data for the RTD environment failed equivalence due to sample mean being too low. The equivalency sample mean (47.082) is 98.74% of the minimum acceptable mean value (47.683). Under the assumption of the modified CV method, the equivalency sample mean passed the test.

The Langley OHC1 strength data for the ETW environment failed equivalence due to both the sample mean and sample minimum being too low. The equivalency sample mean (32.214) is 93.68% of the minimum acceptable mean value (34.387). The equivalency sample minimum (31.638) is 99.61% of the lowest acceptable minimum value (31.763). Under the assumption of the modified CV method, the equivalency sample mean is 95.19% of the lowest acceptable mean value (33.842). The equivalency sample minimum passed the test.

The Marshall OHC1 strength data for the RTD environment failed equivalence due to sample mean being too low. The equivalency sample mean (47.503) is 99.62% of the

Page 40 of 49

167 April 11, 2016 NCP-RP-2016-001 NC

minimum acceptable mean value (47.683). Under the assumption of the modified CV method, the equivalency sample mean passed the test.

The Marshall OHC1 strength data for the ETW environment failed equivalence due to sample mean being too low. The equivalency sample mean (33.186) is 96.51% of the minimum acceptable mean value (34.387). Under the assumption of the modified CV method, the equivalency sample mean is 98.06% of the lowest acceptable mean value (33.842).

HEXCEL 8552 IM7 - NASA Equivalency Test Results and Acceptance Limits Quasi Isotropic Open Hole Compression 60 Strength (OHC1)

30 ksi

0 RTD ETW Environment Glenn Langley Marshall Glenn Averages Langley Averages Marshall Averages RTD Acceptance Limit ETW Acceptance Limit RTD Mod CV Acceptance Limit ETW Mod CV Acceptance Limit Outlier Figure 4-11 OHC1 means, minimums and Acceptance limits

Page 41 of 49

168 April 11, 2016 NCP-RP-2016-001 NC

5. Summary of Results

All the equivalency comparisons are conducted with Type I error probability (α) of 5% in accordance with FAA/DOT/AR-03/19 report and CMH-17 Rev G section 8.4.1. It is common to obtain a few or even several failures in a typical equivalency program involving multiple independent property comparisons. In theory, if the equivalency dataset is truly identical to the qualification dataset, we expect to obtain approximately 5% failures. Since the equivalency test panels were fabricated by a different company, the test panel quality is expected to differ at least marginally; so, we expect to obtain slightly higher failure rates than 5% because the equivalency dataset may not be truly identical to the qualification dataset. However, a failure rate that is significantly higher than 5% is an indication that equivalency should not be assumed and some retesting is justified.

In addition to the frequency of failures, the severity of the failures (i.e. how far away from the pass/fail threshold) and any pattern of failures should be taken into account when making a determination of overall equivalency. Severity of failure can be determined using the graphs accompanying the individual test results. Whether or not a pattern of failures exists is a subjective evaluation to be made by the original equipment manufacturer or certifying agency. The question of how close is close enough is often difficult to answer, and may depend on specific application and purpose of equivalency. NIAR does not make a judgment regarding the overall equivalence; the following information is provided to aid the original equipment manufacturer or certifying agency in making that judgment.

The following computations are based on the assumption that the tests are independent. While the tests are all conducted independently, measurements for strength and modulus are made from a single specimen. Modulus measurements are generally considered to be independent of the strength measurements.

However the computations can be considered conservative. If the tests are not independent and a failure in IPS 0.2% offset strength is correlated with a failure in IPS 5% strain strength, the probability of both failures occurring together should be higher than predicted with the assumption of independence, thus leading to a conservative overall judgment about the material.

Page 42 of 49

169 April 11, 2016 NCP-RP-2016-001 NC

5.1 Failures

The three NASA locations each had 23 different tests conducted but only the modulus tests (8) had sufficient data to meet the CMH-17 Rev G requirements for the individual locations. All strength tests lack sufficient data for the results to be considered conclusive. When the data from the three locations was combined, there was sufficient data for conclusive results for all tests.

Using the modified CV method and examining the tests with sufficient data, the combined dataset had four failures. The main problem area is the compression test results. The OHT1 strength data failed equivalency for both conditions tested.

1. Combined – OHT1 Strength RTD condition failed by 1.6% 2. Combined – OHT1 Strength ETW condition failed by 5.8% 3. Combined – LT modulus ETW condition failed by 0.5% 4. Combined – UNT1 modulus RTD condition failed by 0.2%

5.2 Pass Rate

The combined NASA dataset had four failures out of a 19 tests for a pass rate of 78.95% for NASA overall. If the equivalency samples came from a material with identical properties to the original qualification material and all tests were independent of all other tests, the expected pass rate would be 95%. This equates to 0.95 failures out of 19 tests.

5.3 Probability of Failures

If the equivalency sample came from a material with characteristics identical to the original qualification material and all tests were independent of all other tests, the chance of having four or more failures out of 19 tests is 1.32%.

Figure 5-1 illustrates the probability of getting one or more failures, two or more failures, etc. for a set of 19 independent tests. If the two materials were equivalent, the probability of getting four or more failures is less than 5%. This means that the material could be considered as “not equivalent” with a 95% level of confidence if there were four or more failures out of 19 independent tests.

Page 43 of 49

170 April 11, 2016 NCP-RP-2016-001 NC

Probability of at least x failures in 19 Independent test when materials are equivalent 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 Probability 0.25 0.20 0.15 0.10 0.05 0.00 1 2 3 4 5 6 7 8 9 10 11 12 Number of Failures

Figure 5-1 Probability of Number of Failures

Page 44 of 49

171 April 11, 2016 NCP-RP-2016-001 NC

6. Generic Basis Values Assessment

This material, HEXCEL 8552 IM7, has been used to compute generic basis values, a method designed to take account of the variability introduced by different locations using the same material and methodology. This methods and results were documented in NCAMP Report NCP-RP-2013-015 N/C.

Generic basis values are lower, with lower acceptance criteria, than those computed using the CMH-17 methodology. To determine if a location is able to use these basis values, test results are compared to acceptance criteria for the mean (must be greater than acceptance limit) and standard deviation (must be less than acceptance limit). The sample size of 4 strength specimens is sufficient for this approach. LSBS and Quasi- Isotropic Unnotched tests were not included in the generic basis values analysis, so they cannot be included here.

Since the NASA samples passed the tension strength tests, they also pass the generic acceptance criteria for those tests. Open Hole compression tests results were lower for the NASA samples than the other NCAMP samples of this material. The NASA Glenn Low Vacuum and Slow Ramp tests results were included in the analysis of compression results. The results were similar to the other three NASA samples.

The Generic B-basis values, acceptance criteria and NASA results are given in Table 6-1 for OHC1 tests, in Table 6-2 for LT tests and in Table 6-3 for OHT1 tests. The NASA results are shown graphically along with the complete NCAMP database of HEXCEL 8552 IM7 results for comparison purposes. The results for OHC1 are in Figure 6-1, for LT and OHT1 CTD in Figure 6-2, for LT and OHT1 RTD in Figure 6-3 and for LT and OHT1 ETW results in Figure 6-4.

OHC1 RTD OHC1 ETW Mean Std. Dev. Mean Std. Dev. Generic Acceptance Limit 47.505 2.8052 33.722 2.5744 Glenn 45.792 0.973 31.767 0.444 Glenn: Low Vacuum 47.416 1.575 32.678 1.193 Glenn: Slow Ramp 46.011 1.461 34.148 1.626 Langley 47.082 1.379 32.678 0.766 Marshall 47.503 1.193 33.186 0.910 Table 6-1 OHC1 Generic Basis Values Acceptance Criteria and NASA test results

Page 45 of 49

172 April 11, 2016 NCP-RP-2016-001 NC

Hexcel 8552 IM7 Unidirectional Prepreg Open Hole Compression Strength RTD and ETW Conditions with Basis values and Acceptance Limits 40

34 ETW Condition (ksi)ETW 33

30 44 45 46 47 48 49 50 51 52 53 54 RTD Condition (ksi)

Qualification Unit Values Equivalency Unit Values Generic Acceptance Region for OHC RTD and ETW Conditions Generic Equivalence Region for OHC RTD and ETW Condition Qualification B-basis Qualification Acceptance Limit for Mean Pooled B-basis (ANOVA) Pooled (ANOVA) Acceptance Limit for Mean Generic B-basis NASA Figure 6-1 NASA OHC1 results with NCAMP Generic Criteria and Basis values

LT CTD LT RTD LT ETW Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Generic Acceptance Limit 285.24 30.05 294.68 42.21 254.31 45.92 Glenn 363.66 12.88 388.29 6.43 356.37 11.53 Langley 368.17 11.95 381.73 14.21 358.08 13.23 Marshall 358.03 6.84 397.14 10.80 366.51 12.90 Table 6-2 LT Generic Basis Values Acceptance Criteria and NASA test results

OHT1 CTD OHT1 RTD OHT1 ETW Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Generic Acceptance Limit 43.81 4.11 49.59 3.83 59.35 3.84 Glenn 64.58 1.47 65.99 1.85 69.24 4.80 Langley 63.56 1.76 62.41 1.12 68.99 1.68 Marshall 62.63 1.35 64.16 1.70 69.36 2.37

Table 6-3 OHT1 Generic Basis Values Acceptance Criteria and NASA test results

Page 46 of 49

173 April 11, 2016 NCP-RP-2016-001 NC

Hexcel 8552 IM7 Unidirectional Prepreg Longitudinal Tension Strength and Open Hole Tension Strength CTD Condition with Basis values and Acceptance Limits 80

Open Hole Tensino Strength (ksi) 45

35 250 270 290 310 330 350 370 390 410 430 450 Longitudinal Tension Strength (ksi) Qualification Unit Values Equivalency Unit Values Generic Acceptance Region for LT and OHT CTD Condition Generic Equivalence Region for LT and OHT CTD Condition Qualification B-basis Qualification Acceptance Limit for Mean Pooled B-basis Pooled Sample Acceptance Limit for Mean Generic B-basis NASA Figure 6-2 NASA LT & OHT1 CTD results with NCAMP Generic Criteria and Basis values

Hexcel 8552 IM7 Unidirectional Prepreg Longitundinal Tension Strength and Open Hole Tension Strength RTD Condition with Basis values and Acceptance Limits 75

50 Open Hole Tension Strength (ksi) 45

40 250 270 290 310 330 350 370 390 410 430 450 Longitudinal Tension Strength (ksi) Qualification Unit Values Equivalency Unit Values Generic Equivalence Region Generic Acceptance Region Qualification B-basis Qualification Acceptance Limit for Mean Pooled B-basis Pooled Sample Acceptance Limit for Mean Generic B-basis Generic Acceptable but not Equivalent Region NASA Figure 6-3 NASA LT & OHT1 RTD results with NCAMP Generic Criteria and Basis values

Page 47 of 49

174 April 11, 2016 NCP-RP-2016-001 NC

Hexcel 8552 IM7 Unidirectional Prepreg Longitudinal Tension Strength and Open Hole Tension Strength ETW Condition with Basis values and Acceptance Limits 80

60 Open Hole Tension Strength (ksi) 55

50 230 280 330 380 430 Longitudinal Tension Strength (ksi) Qualification Unit Values Equivalency Unit Values Generic Acceptance Region for LT and OHT ETW Condition Generic Equivalence Region for LT and OHT ETW Condition Qualification B-basis Qualification Acceptance Limit for Mean Pooled B-basis Pooled Sample Acceptance Limit for Mean Generic B-basis NASA Figure 6-4 NASA LT & OHT1 ETW results with NCAMP Generic Criteria and Basis values

Page 48 of 49

175 April 11, 2016 NCP-RP-2016-001 NC

7. References

1. CMH-17 Rev G, Volume 1, 2012. SAE International, 400 Commonwealth Drive, Warrendale, PA 15096 2. John Tomblin, Yeow C. Ng, and K. Suresh Raju, “Material Qualification and Equivalency for polymer Matrix Composite Material Systems: Updated Procedure”, National Technical Information Service (NTIS), Springfield, Virginia 22161 3. Vangel, Mark, "Lot Acceptance and Compliance Testing Using the Sample Mean and an Extremum", Technometrics, Vol 44, NO. 3, August 2002, pp. 242-249

Page 49 of 49

176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 APPENDIX C––TECHNOLOGY READINESS LEVEL ASSESSMENT

The CEUS project team assessed the TRL and, in the case of manufacturing, the manufacturing readiness assessment (MRL) of 8.4-m-diameter composite dry structures based on the limited accomplishments of the project. Table 30 reflects the results of that assessment. Overall, there are several individual technology areas in the TRL 3–5 range within structures and manufacturing that need to be matured to enable the overall composites capability for future space vehicles.

Table 30. Assessment of TRL/MRL of 8.4-m-diameter composites technologies.

Historically Today Human-rated composites Prior experience up to 3.6 m diameter TRL not ratable for 8.4 m composites 8.4-m-diameter forward and aft Improved design/analysis/manufacturability experience skirts TRL 4 Automated manufacturing Advancement in low-cost tooling to support larger diameter (8.4 m) segmented construction MRL 5–6 OOA joints Designed/analyzed OOA joints at 8.4 m diameter TRL 3 Design database Utilized existing material database on CEUS and demonstrated equivalency TRL 8

Several months prior to the beginning of the CEUS project, an MSFC-LaRC-GRC team examined the perceived impediments to the utilization of composites in flight structures within the Agency. Table 31 is a quick look at where the CEUS project team believes we are in addressing some of those impediments. Composites are already being used in flight structures at dimensions of up to 5 m diameter, and are ready for the next large step in technology maturation and infusion into an 8.4-m, human-rated flight hardware path.

237 Table 31. Assessment of composite technologies versus perceived impediment.

Perceived Impediment (2013) Status Today Affordability means limiting risk by using what we CEUS would have demonstrated design/analysis/manufactur- already know (heritage hardware). ing/test of new hardware with minimal risk. Considerable work remains to mature the technology. ‘Composites’ means funding costly materials CEUS utilized existing database for an established material development programs. with process modifications—much less cost and risk. NASA design and certification standards are driving Standards are tailorable. An aggressive program can take a ‘stacking of conservatism’ that is limiting the perfor- advantage of such tailoring and reduce cost and schedule while mance gains offered by composite structures. (Does increasing performance. not allow for an affordable approach.) Composites will mean developing costly new CEUS developed low-cost tooling and utilized existing facilities manufacturing processes and tooling and negate the to keep infrastructure costs low. advantages of common tooling. Meeting the damage tolerance requirements for CEUS developed designs to account for damage; utilized composites will introduce significant technical/ damage-based material properties to significantly reduce risks. schedule/cost risks.

238 APPENDIX D––DESIGN DATA SHEETS AND DRAWINGS

Design data sheets for the STA and Pathfinder and two drawings for the STA are given in this appendix.

239 STA_DDS_mod_3-29-16.doc

Composites Technology for Exploration (CTE) Structural Test Article Design Data Sheet

INTENT: This document will provide specific design criteria for the CTE STA geometry, loads, and overall description. Analysis guidance for composites will be documented in Composites for Exploration Upper Stage – CEUS Structural Analysis Ground Rules and Assumptions.

240 Composites Technology for Exploration STA

Date: 3/29/2016

Page 1 of 4

Layout

13 ft (156”) Not to Exceed Length 12 ft (144”) Composite Length

36” Penetration

Outer Diameter = 331.2”

Construction:

• 8 sandwich composite panels to be joined with out-of-autoclave, out-of-oven vertical joints. • Metallic end rings mechanically attached to the existing simulators with existing bolt pattern.

Manufacturing:

• Use automated fiber placement machine to lay up face sheets. • Autoclave curing of the facesheets. • Assembly fixture to support out-of-autoclave, out-of-oven vertical joints.

241 Composites Technology for Exploration STA

Date: 3/29/2016

Page 2 of 4

Materials

Composite:

• Intermediate modulus/toughened epoxy skins

Adhesive:

• Epoxy

Potting:

• Epoxy Liquid shim between the metal fitting and the composite surface

Core:

• Aluminum honeycomb core • Higher density composite inserts (plugs) to reinforce base and lifting points

End Rings:

• Separable Clevis design • Aluminum • Joined by aluminum splice plates

Vertical Joints:

• TBD

End Rings, Fwd and Aft (Notional)

242 Composites Technology for Exploration STA

Date: 3/29/2016

Page 3 of 4

Vehicle Line Loads

Coupled loads analysis: EV31 DAC0 Block 1B Cargo and Crew Configurations.

Requirement to test to a representative flight load case. Bin 5 was chosen for the STA to facilitate using the existing External Tank test simulators without structural modification.

Compression for Ascent Bin 5

P (lb) M (in-lb) V (lb) Cylinder fwd ITAR Information ITAR Information ITAR Information Removed Removed Removed Cylinder aft ITAR Information ITAR Information ITAR Information Removed Removed Removed

156.0” CTE STA max (abbreviated cylinder)

Baseline USA2

243 Composites Technology for Exploration STA

Date: 3/29/2016

Page 4 of 4

Venting Pressures (Cylinder):

• Burst Pressure: 0 psi • Crush Pressure: 0 psi

Knockdown Factor (KDF):

• Discrepancies between analytically and empirically derived buckling load capability are due in part to the differences between idealized model geometry and the physical structure. “Knockdown factors” (correlation coefficients) are used to adjust predicted values to account for these differences. Typical knockdown factors for thin-walled circular cylinders are listed in NASA SP-8007, Buckling of Thin-Walled Circular Cylinders.

Design Details to Accommodate:

• STA will include one 36” penetration positioned at least 1D from top of panel. • Handling Provisions needed for EM

Temperature: Ambient

Design details above and overall geometry will be documented in layout drawing (SMDB-CTE-0003).

244 Pathfinder_DDS_mod_3-29-16.doc

Composites Technologies for Exploration (CTE) Pathfinder Design Data Sheet

INTENT: This document will provide specific design criteria for the CTE Pathfinder geometry, loads, and overall description. Analysis guidance for composites will be documented in Composites for Exploration Upper Stage – CEUS Structural Analysis Ground Rules and Assumptions.

245 Composites Technologies for Exploration Pathfinder

Date: 03/29/2016

Page 1 of 7 Layout

Per SPIE Baseline:

Construction:

• Honeycomb or foam sandwich composite panels. • Vertical separation joints and horizontal separation joints. • Metallic clevis type end rings joined by splice plates

246 Composites Technologies for Exploration Pathfinder

Date: 03/29/2016

Page 2 of 7 Notional Dynamic Envelope (Keep-out Zone):

• From drawing 150601MRA002, USA2-1000-OML

247 Composites Technologies for Exploration Pathfinder

Date: 03/29/2016

Page 3 of 7 Trade Studies

Multiple trade studies conducted to explore different configurations:

• Geometry Study:

Two Part: Cone Alternate Geometry and Cylinder

• Metallic Acreage vs. Composite Acreage Study

• Core Material Study: Aluminum Honeycomb or Foam

• Core Density Study

• Knockdown Factor Study

• Vertical Separation Joint Study

• Discontinuity Factor Sensitivity Study

248 Composites Technologies for Exploration Pathfinder

Date: 03/29/2016

Page 4 of 7 Vehicle Line Loads

Coupled loads analysis: EV31 DAC0 Block 1B Cargo and Crew Configurations.

Consider DAC0 to be the final verification loads cycle for the Pathfinder design.

249 Composites Technologies for Exploration Pathfinder

Date: 03/29/2016

Page 5 of 7 Venting Pressures

ΔPskin Results by Bin (Enveloped Qmax, Qmin):

Mach Bin Definitions:

250 Composites Technologies for Exploration Pathfinder

Date: 03/29/2016

Page 6 of 7 Knockdown Factor (KDF):

• Typical knockdown factors are listed in NASA SP-8007, Buckling of Thin-Walled Circular Cylinders, NASA SP-8019 Buckling of Thin-Walled Truncated Cones, and NASA SP-8032 Buckling of Thin-Walled Doubly Curved Shells

Stiffness Requirement:

• From DAC1 MSFC Loads model

Free Free Frequency Mode # (Hz) 1 Removed 2 Removed 3 Removed 4 Removed 5 Removed 6 Removed 7 Removed 8 Removed 9 Removed 10 Removed

Mass (lbm) Removed Stiffness (lb/in) Axial Removed Shear 1 Removed Shear 2 Removed

251 Composites Technologies for Exploration Pathfinder

Date: 03/29/2016

Page 7 of 7 Temperature:

• Transient temperature profiles will be developed for hot/cold pre-launch DSNE terrestrial environments (e.g., roll-out, on-pad, tanking) as well as ascent aero-thermodynamic environments. o Initial case will assume no TPS o Aero-thermal heating data not currently available for USA2 in the Block 1B DAC-0 database, will coordinate with EV33 for reasonable aero-heating assumptions for preliminary assessment o Initial results May 1, 2016

252 ENGINEER_SMDB-CTE-0002.pdf

253 ENGINEER_SMDB-CTE-0003.pdf

254 255 APPENDIX E––MANUFACTURING

Appendix E contains the Thermographic Inspection Report.

256 Thermographic Inspection Report Work Order 2016-0063

Prepared For Prepared By James Walker Scott Ragasa NASA MSFC METTS [email protected] [email protected] (256) 961-1784 (256) 544-3935

Specimen Information

Project CEUS Serial Number CEUS_001A, _001B, _006 Surface Preparation None Special Handling None

Inspection Equipment

Infrared Camera FLIR SC6000 Lens 25 mm Heating Method Flash Lamps Hood Configuration Small FOV

Inspection Settings

Capture Software EchoTherm 8 Image Size 640 x 512 Capture Frequency 30 Hz Capture Duration 9.2 seconds Flash Duration 30 milliseconds Flash Delay 0 milliseconds Flash Frame 10 TSR Skip Frames 1

257 February 19, 2016 WR: 2016-0063 Revision A

Remarks

 A thermography inspection was performed to determine if there were any initial defects in the set of panels prior to mechanical testing.  Due to the low emissivity of the panel surfaces, only derivative processing of the images was used.  One indication was found in panel CEUS_006, Front, Location J02.

Page 2 of 9

258 February 19, 2016 WR: 2016-0063 Revision A

CEUS_006 – Front – Location J02

Raw, Frame 100, ROI Adjusted T-t Plot

The logarithmic time versus temperature plot shows the area of interest (red) deviates slightly from the background (blue).

The regular shape suggests it is likely FOD – tape or backing material.

Page 3 of 9

259

–

4 9

260

–

5 9

261

–

6 9

262

–

7 9

263

–

8 9

264

–

9 9

265 REFERENCES

1. NASA MPR 7120.1, Rev. G, “MSFC Engineering and Program/Project Management Requirements,” August 26, 2014.

2. NASA-STD-5019, “Fracture Control Requirements for Spaceflight Hardware,” NASA, Washington, DC, 44 pp., 2008.

3. MSFC-RQMT-3479, “Fracture Control Requirements for Composite and Bonded Vehicle and Payload Structures,” NASA Marshall Space Flight Center, Huntsville, AL, 50 pp., June 2006.

4. Clarkson, E.: “Hexcel 8552 IM7 Unidirectional Prepreg 190 gsm and 35% RC Qualification Statistical Analysis Report,” NCP-RP-2009-028, Rev N/C, National Institute for Aviation Research, Wichita State University, Wichita, pp. 104, January 2011.

5. CMH-17, Rev. G, “Polymer Matrix Composites: Guidelines for Characterization of Structural Materials,” SAE International, Warrendale, PA, 711 pp., 2012.

6. Ng, Y.; and Tomblin, J.: “Fabrication of NMS 128 Qualification, Equivalency, and Acceptance Test Panels (for Hexcel 8552 and 8552S prepregs),” National Institute for Aviation Research, NCAMP Process Specification, NPS 81228, Rev. B, 2011.

7. Wu, K.C.; Stewart, B.K.; and Martin, R.A.: “ISAAC––A Testbed for Advanced Composite Research,” Paper presented at The American Society for Composites 29th Annual Technical Conference, San Diego, CA, September 8–10, 2014.

8. Jeffries, K.A.: “Enhanced Robotic Automated Fiber Placement with Accurate Robot Technology and Modular Fiber Placement Head,” SAE Int. J. Aerosp., Vol. 6, No. 2, doi: 10.4271/2013-01-2290, September 17, 2013.

9. Krivanek, T.M.; and Yount, B.C.: “Composite Payload Fairing Structural Architecture Assess- ment and Selection,” Technical Report E-18094, NASA Glenn Research Center, Cleveland, OH, 15 pp., May 2012. 10. NASA-STD-6016, “Standard Materials and Processes Requirement for Spacecraft,” Johnson Space Center, Houston, TX, September 25, 2009.

266 11. Pineda, E.J.; Myers, D.E.; Bednarcyk, B.A.; and Krivanek, T.M.: “A Numerical Study on the Effect of Facesheet-Core Disbonds on the Buckling Load of Curved Honeycomb Sandwich Panels,” Paper presented at the American Society for Composites 30th Technical Conference, Lansing, MI, September 28–30, 2015.

12. Rinker, M.K.; Krueger, R.; and Ratcliffe, J.: “Analysis of an Aircraft Honeycomb Sandwich Panel with Circular Face Sheet/Core Disbond Subjected to Ground-Air Pressurization,” NASA/CR—2013–217974, NASA Langley Research Center, Hampton, VA, 32 pp., March 2013.

13. Glaessgen, E.H.; Reeder, J.R.; Sleight, D.W.; et al.: “Debonding Failure of Sandwich- Composite Cryogenic Fuel Tank with Internal Core Pressure,” Journal of Spacecraft and Rockets, Vol. 42, No. 4, pp. 613–627, 2005.

14. Ratcliffe, J.: “Sizing Single Cantilever Beam Specimens for Characterizating Facesheet/Core Peel Debonding in Sandwich Structure,” ICSS 9, G. Ravichandran, Editor, 9th International Conference on Sandwich Structures, Pasadena, CA, June 14–16, 2010, 13 pp., 2011.

15. Reeder, J.R.; Song, K.; Chunchu, P.B.; and Ambur, D.R.: “Postbuckling and Growth of Delaminations in Composite Plates Subjected to Axial Compression,” Paper presented at the 43rd AIAA/ASME/ASCE/AHS Structures, Structural Dynamics, and Materials Conference, Denver, CO, April 22–25, 2002.

16. Ural, A.Z.; Zehnder, A.T.; and Ingraffea, A.R.: “Fracture Mechanics Approach to Facesheet Delamination in Honeycomb: Measurement of Energy Release Rate of the Adhesive Bond,” Engineering Fracture Mechanics, Vol. 70, Issue 1, pp. 93–103, 2003.

17. Turon, A.D.; Davila, C.G.; Camanho, P.P.; and Costa, J.: “An Engineering Solution for Solving Mesh Size Effects in the Simulation of Delamination Using Cohesive Zone Models,” Engineering Fracture Mechanics, Vol. 74, pp. 1665–1682, 2007.

18. Krueger, R.: “A Summary of Benchmark Examples to Assess the Performance of Quasi-Static Delamination Propagation Prediction Capabilities in Finite Element Codes,” Journal of Composite Materials, Vol. 49, pp. 3297–3316, November 2015.

19. Thévenin, R.: “Airbus Composite Structures: Perspectives on Safe Maintenance Practice,” Paper Presented at the Commercial Aircraft Committee Meeting and Related Industry/FAA/ EASA Workshop, Amsterdam, 43 pp., May 7–11, 2007.

267 Form Approved REPORT DOCUMENTATION PAGE OMB No. 0704-0188 The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operation and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.

1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To) 01–12–2016 Technical Memorandum 4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER

Composites for Exploration Upper Stage 5b. GRANT NUMBER

5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

J.C. Fikes, J.R. Jackson, S.W. Richardson, A.D. Thomas, 5e. TASK NUMBER T.O. Mann,* and S.G. Miller** 5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION George C. Marshall Space Flight Center REPORT NUMBER Huntsville, AL 35812 M–1422

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITOR’S ACRONYM(S) National Aeronautics and Space Administration NASA Washington, DC 20546–0001 11. SPONSORING/MONITORING REPORT NUMBER NASA/TM—2016–219433

12. DISTRIBUTION/AVAILABILITY STATEMENT Unclassified-Unlimited Subject Category 39 Availability: NASA STI Information Desk (757–864–9658)

13. SUPPLEMENTARY NOTES Prepared by the Technology Development and Transfer Office, Science and Technology Office *Langley Research Center, **Glenn Research Center

14. ABSTRACT The primary objective of the Composites for Exploration Upper Stage project was to mature the technology readiness of composite dry structures at diameters that are suitable for Space Launch System (SLS) Exploration Upper Stage and other in-space applications. SLS studies have shown that composites provide important programmatic enhancements, including increased capability and accelerated expansion of exploration and science mission objectives. Reducing the weight and cost of these structures will provide a significant benefit, perhaps even an enabling benefit, to an upper stage. This initiative was a cooperative effort between the Space Technology Mission Directorate and the Human Exploration and Operations Mission Directorate, including multiple NASA Centers.

15. SUBJECT TERMS exploration, Upper Stage, Ares, Space Launch System

16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF ABSTRACT 18. NUMBER OF 19a. NAME OF RESPONSIBLE PERSON a. REPORT b. ABSTRACT c. THIS PAGE PAGES STI Help Desk at email: [email protected] U U U UU 294 19b. TELEPHONE NUMBER (Include area code) STI Help Desk at: 757–864–9658

Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std. Z39-18

268

National Aeronautics and NASA/TM—2016–XXXX Space Administration IS02 George C. Marshall Space Flight Center Huntsville, Alabama 35812

Closeout Report for Composites for Exploration Upper Stage

J.C. Fikes, J.R. Jackson, S.W. Richardson, and A.D. Thomas Marshall Space Flight Center, Huntsville, Alabama

T.O. Mann Langley Research Center, Hampton, Virginia

S.G. Miller Glenn Research Center, Cleveland, Ohio

October 2016